Research ArticleCancer

Dimethyl fumarate potentiates oncolytic virotherapy through NF-κB inhibition

See allHide authors and affiliations

Science Translational Medicine  24 Jan 2018:
Vol. 10, Issue 425, eaao1613
DOI: 10.1126/scitranslmed.aao1613

A virus’s little helper

Oncolytic viruses, which kill cancer cells, can offer an effective and versatile approach for treating cancer. However, they need to reach tumor cells and get inside them to achieve a therapeutic effect, and this does not always happen. Selman et al. identified a promising solution for this problem by combining oncolytic vesicular stomatitis virus with dimethyl fumarate, a small-molecule drug that is already in use for some nonmalignant disorders and may also have direct anticancer effects. Dimethyl fumarate promoted viral infection of cancer cells, and the combined treatment was effective in multiple cancer models, including those that did not respond to virus or drug treatment alone.

Abstract

Resistance to oncolytic virotherapy is frequently associated with failure of tumor cells to get infected by the virus. Dimethyl fumarate (DMF), a common treatment for psoriasis and multiple sclerosis, also has anticancer properties. We show that DMF and various fumaric and maleic acid esters (FMAEs) enhance viral infection of cancer cell lines as well as human tumor biopsies with several oncolytic viruses (OVs), improving therapeutic outcomes in resistant syngeneic and xenograft tumor models. This results in durable responses, even in models otherwise refractory to OV and drug monotherapies. The ability of DMF to enhance viral spread results from its ability to inhibit type I interferon (IFN) production and response, which is associated with its blockade of nuclear translocation of the transcription factor nuclear factor κB (NF-κB). This study demonstrates that unconventional application of U.S. Food and Drug Administration–approved drugs and biological agents can result in improved anticancer therapeutic outcomes.

INTRODUCTION

Deregulated metabolism and defective innate immune response are common characteristics of transformed cells (1). This makes it possible to genetically engineer or select oncolytic viruses (OVs), which specifically infect and kill tumor cells without harming normal tissues. OVs can be generated from various viral backbones using diverse genetic approaches and provide capacity for expressing therapeutic or imaging transgenes. Some of the early OV candidates have finally made their way to the clinic with the approval of talimogene laherparepvec (or T-Vec) based on herpes simplex virus 1 (HSV-1) for the treatment of advanced melanoma (2).

Preclinical and clinical trials have showcased the heterogeneity in the therapeutic response to OV treatment, in which a subset of patient tumors is refractory to infection with OVs (36). The identification of pharmacological agents that can functionally enhance OVs to improve therapeutic benefit has been an area of intense investigation and has been recognized as critical to maximize the therapeutic impact of OVs in the clinic (6). Combination strategies with checkpoint blockade have shown promise recently (79) but do not address the frequent resistance of tumors to viral infection. To this end, strategies aiming to suppress antiviral immune responses, for example, using immunosuppressants such as cyclophosphamide, rapamycin, and histone deacetylase inhibitors, have shown promise in vitro but variable effects in animal models (1013). Combinations of OVs with cyclophosphamide, in particular, are under clinical evaluation, with favorable safety profiles reported thus far (NCT01598129).

Fumaric acid esters (FAEs) are a class of compounds with known anti-inflammatory and neuroprotective effects (1416). The mechanisms involved have yet to be fully elucidated but are thought to be mediated through the activation of the antioxidative transcription factor nuclear factor (erythroid-derived 2)–like 2 (NRF2) pathway (15), the inhibition of nuclear factor κB (NF-κB) (17), as well as the functional depletion of glutathione (GSH) (18, 19). FAEs (marketed as Fumaderm and Psorinovo) were first approved as a treatment for psoriasis in Germany. Recently, dimethyl fumarate (DMF), an FAE marketed as Tecfidera, was approved by the U.S. Food and Drug Administration and the European Medicines Agency for the treatment of relapsing forms of multiple sclerosis (MS) and relapsing-remitting MS (20). Clinical studies on the long-term use of DMF have not revealed any severe long-term adverse effects (2022). Recent reports suggest that DMF has anticancer potential, shown to suppress tumor growth and metastasis (2328) in addition to sensitizing tumors to chemotherapy (26, 29). Furthermore, DMF is currently under clinical evaluation for the treatment of chronic lymphocytic leukemia and cutaneous T cell lymphoma (NCT02546440 and NCT02784834). Given this and the documented positive effect of diverse immunosuppressants in combination with OVs, we set out to explore the possibility of using DMF in combination with oncolytic virotherapy in vitro and in vivo.

RESULTS

DMF enhances tumor-specific viral spread and oncolysis in vitro

Little is known about the effect of DMF on viral infection. We first examined the impact of DMF (structure displayed in Fig. 1A) on the growth of oncolytic vesicular stomatitis virus (VSVΔ51). Notably, VSV and closely related Maraba are currently undergoing clinical evaluation but face resistance to infection in about a third of cancers (5, 30). 786-0 renal carcinoma cells are highly refractory to VSVΔ51 infection; however, a 4-hour pretreatment with DMF at doses between 50 and 200 μm increased VSVΔ51 virus growth at a low multiplicity of infection (MOI) by more than 100-fold (Fig. 1B). Virus-encoded green fluorescent protein (GFP) expression was also greatly enhanced, as visualized by microscopy (Fig. 1C). This effect was observed with pretreatment times up to 24 hours before infection, and after infection up to 8 hours (fig. S1A). More broadly, DMF robustly enhanced infection in a panel of human and murine cancer cell lines (sarcoma, osteosarcoma, breast, colon, melanoma, and ovarian) with a wide range of sensitivity to VSVΔ51 (Fig. 1D). DMF also improved viral output of Sindbis virus and oncolytic herpes virus (HSV-1 mutant N212) after treatment in 786-0 cells (Fig. 1, E and F). Furthermore, we also tested the effect of DMF on adenoviruses, which, in addition to oncolytic virotherapy, have been extensively used as gene therapy vectors (31). Pretreatment of human lung adenocarcinoma A549 cells with DMF increased luciferase transgene expression of a nonreplicating adenovirus, E1A-deleted Ad5, over the course of 7 days by up to 20-fold (Fig. 1G).

Fig. 1 DMF promotes viral infection.

(A) Structure of DMF. (B) VSVΔ51-resistant human renal cancer cell line 786-0 was pretreated with DMF or left untreated for 4 hours and subsequently infected with VSVΔ51 (MOI, 0.01). Corresponding viral titers were determined 48 hours after infection from supernatants [n = 3; mean ± SD; *P < 0.05, ***P < 0.001, one-way analysis of variance (ANOVA), as compared to the untreated condition]. PFU, plaque-forming units. (C and D) Various human and murine cell lines were pretreated with DMF (150 to 250 μM) or left untreated for 4 hours and subsequently infected with VSVΔ51 (MOI, 0.01). Twenty-four hours after infection, fluorescence images were taken from the infected cancer cells, as shown in (C). Corresponding viral titers were determined 48 hours after infection from supernatants, as shown in (D) (n = 3 to 4; mean ± SD; P < 0.05, two-tailed t test, as compared to the untreated counterpart for each cell line). (E and F) 786-0 cells were pretreated with DMF (150 μM) or left untreated for 4 hours and subsequently infected with (E) Sindbis virus (MOI, 10) or (F) HSV-1 (MOI, 0.01). Corresponding viral titers in supernatants were determined 48 hours after infection (n = 3; mean ± SD; **P < 0.01, ***P < 0.001, two-tailed t test, as compared to the untreated counterpart). (G) Human A549 cells were pretreated as in (B) and infected with an adenovirus expressing firefly luciferase (Ad5) at an MOI of 1. Luciferase activity was measured over the course of 7 days. Results are represented as relative light units (RLU), and background is indicated by a black line (n = 3; mean ± SD; P < 0.05, two-way ANOVA from day 2 to day 5). (H) Multistep (MOI, 0.001 and 0.01) and single-step (MOI, 3) growth curves. 786-0 cells were pretreated with DMF and infected with VSVΔ51 at an MOI of 0.001, 0.01, or 3; supernatants were titered by plaque assay (n = 3; mean ± SD). (I) 786-0 cells were pretreated with DMF for 4 hours and infected with VSVΔ51 (MOI, 0.0001). An agarose overlay was added after 1 hour of infection. Fluorescence microscopy of a representative plaque 48 hours after infection is shown. The complete dose range is presented in fig. S1B. Corresponding images of Coomassie blue stain of the full wells and the graph of average plaque diameters illustrate the enhancement of the plaque diameters in the presence of DMF (n = 20; horizontal lines indicate means; *P < 0.05, ***P < 0.001, one-way ANOVA, as compared to the mock-treated counterpart). (J) 786-0, CT26WT, and B16F10 cell lines were pretreated and infected as in (B). Cell viability was assayed 48 hours after infection. Results were normalized to the average of the values obtained for the corresponding uninfected, untreated cells (n = 8; mean ± SD; ***P < 0.001, one-way ANOVA, as compared to VSVΔ51 condition).

Single- and multistep growth curves performed in 786-0 cells revealed that DMF enhanced the viral output of VSVΔ51 when using a low MOI, but not when using a high MOI (Fig. 1H), indicating that DMF inhibits mechanisms that impinge on viral spread rather than virus output per infected cell. To further explore the effect of DMF on virus spreading, we infected 786-0 cell monolayers with VSVΔ51 and overlaid them with agarose to generate defined plaques of virus replication foci. DMF substantially increased the average plaque diameter of VSVΔ51 on fluorescence imaging and Coomassie blue staining of infected cell monolayers (Fig. 1I and fig. S1B). To further assess the oncolytic effect of VSVΔ51 in the presence of DMF, we pretreated cancer cells with DMF before infection with VSVΔ51 at a low MOI, and cell viability was assessed with the metabolic dye alamarBlue 48 hours after infection. Combined treatment with DMF and VSVΔ51 resulted in a decrease in cell viability in human 786-0 as well as murine CT26WT and B16F10 cells (Fig. 1J).

We further assessed the ability of DMF to enhance VSVΔ51 infection ex vivo in mouse-derived tissues. Tumor cores from mice subcutaneously implanted with CT26WT murine colon cancer cells or B16F10 murine melanoma cells, as well as cores from normal lung, muscle, and spleen, were collected and subsequently infected with VSVΔ51-GFP in the presence or absence of 150 μM DMF. DMF robustly increased the growth of the virus in CT26WT and B16F10 cores by 31- and 13-fold, respectively, but did not increase virus growth in normal tissue cores (Fig. 2A). Furthermore, DMF increased VSVΔ51 infection by >10-fold ex vivo in primary human melanoma, lung, prostate, and ovarian tumor samples, as observed by plaque assay and microscopy (Fig. 2, B and C), as well as in various melanoma and ovarian patient-derived cancer cell lines (Fig. 2D). Similar to the murine normal tissue cores, DMF did not promote virus growth in human normal lung and muscle tissues from various patients (Fig. 2E).

Fig. 2 DMF enhances infection ex vivo and in human clinical samples.

(A) CT26WT and B16F10 tumors were grown subcutaneously in BALB/c and C57BL/6 mice, respectively, and excised. BALB/c and C57BL/6 mouse spleens, muscles, lungs, and brain tissue were also collected and cored. Tumor and normal tissue cores were pretreated with 150 μM DMF for 4 hours and subsequently infected with 1 × 104 PFU of oncolytic VSVΔ51 expressing GFP. Twenty-four hours after infection, fluorescence images were acquired for the tumor or normal tissue cores. Representative images from each triplicate set are shown in the upper panel. Viral titers from supernatant were determined 48 hours after infection and are shown in the lower panel (n = 3 to 4; mean ± SD; ns, not significant; *P < 0.05, two-tailed t test, as compared to the untreated counterpart). (B) Indicated human tumor tissues were treated with DMF for 4 hours and subsequently infected with 1 × 104 PFU of VSVΔ51 expressing GFP. Viral titers were determined 48 hours after infection (n = 3 to 4; mean ± SD; *P < 0.05, **P < 0.01, two-tailed t test, as compared to the untreated counterpart). (C) Representative fluorescence and bright-field images are shown for patient 8. (D) Patient-derived cell lines were treated with DMF for 4 hours and subsequently infected with an MOI of 0.01 of oncolytic VSVΔ51 expressing GFP. Corresponding viral titers were determined 48 hours after infection (n = 3; mean ± SD; *P < 0.05, two-tailed t test, as compared to the untreated counterpart). (E) Human normal tissue was treated as in (B) (n = 3 to 4; mean ± SD; two-tailed t test).

In addition to DMF, various fumaric and maleic acid esters (FMAEs) exhibit anti-inflammatory and immunomodulatory properties (32). We therefore tested whether other FAEs and their cis- and trans-isoforms (maleic acid esters) (Fig. 3, A and B) have a similar impact to DMF on viral infection of cancer cells. The cell-permeable FAE diethyl fumarate (DEF) and maleic acid esters dimethyl maleate (DMM) and diethyl maleate (DEM) robustly enhanced VSVΔ51 infection, spread, and oncolysis in 786-0 and CT26WT cells in vitro (Fig. 3C and figs. S2 and S3). Enhanced infection was also observed with FMAEs in CT26WT tumor cores infected ex vivo (Fig. 3D). DMF is rapidly hydrolyzed to monomethyl fumarate (MMF) by esterases in vivo and subsequently to FA (as displayed in Fig. 3B) (33). MMF is thought to be the active metabolite of DMF in the treatment of MS (33). In 786-0 cells, MMF also enhanced infection of VSVΔ51 to a similar extent as DMF, albeit at higher effective doses. Fumaric acid (FA), in contrast with the cell-permeable esters MMF and DMF, had no impact on viral growth (Fig. 3C and fig. S2). Together, our data indicate that DMF and other cell-permeable FMAEs can markedly enhance the spread of OVs in both mouse and human cell lines and cancer tissue explants.

Fig. 3 FMAEs promote infection by VSVΔ51.

(A) Structures of FAEs (DEF) and maleic acid esters (DEM and DMM). (B) Metabolism of DMF. DMF is hydrolyzed into monomethyl fumarate (MMF), which in turn is metabolized into FA. FA subsequently enters the Krebs cycle. (C) 786-0 cells or (D) CT26WT ex vivo tumor cores were pretreated with various FMAEs and analogs for 4 hours and subsequently infected with oncolytic VSVΔ51 expressing GFP at (C) an MOI of 0.01 or (D) 1 × 104 PFU. Twenty-four hours after infection, we obtained fluorescence images of the infected 786-0 cells or CT26WT tumor cores. Corresponding viral titers were determined from supernatants 48 hours after infection (n = 3; mean ± SD; *P < 0.05, **P < 0.01, ***P < 0.001, one-way ANOVA, as compared to the untreated counterpart).

DMF improves therapeutic efficacy of oncolytic rhabdoviruses

Because DMF is a clinically approved drug and it broadly and robustly enhanced the growth and activity of VSVΔ51 in vitro in both human and mouse tumor explants, and did so preferentially in tumors opposed to normal tissues, we next evaluated the potential therapeutic benefit of combining DMF with oncolytic VSVΔ51 in vivo. To this end, we used both syngeneic and xenograft mouse tumor models, in which VSVΔ51 is ineffective as a monotherapy (3437). Mouse CT26WT and B16F10 and human colon cancer HT29 cells were grown subcutaneously in BALB/c, C57BL/6, or nude mice, respectively. Mice were injected intratumorally with DMF for 4 hours and subsequently infected with VSVΔ51 expressing luciferase. With the exception of the B16F10 model, DMF enhanced virus-associated luciferase gene expression specifically in tumors 24 hours after the first injection of virus, as assessed using an in vivo imaging system (IVIS) (Fig. 4, A and B). In all three models, the combination therapy considerably delayed tumor progression (Fig. 4C and table S1) and significantly prolonged survival compared with either monotherapy (combination therapy compared to VSVΔ51 alone; CT26WT, P = 0.0008; B16F10, P = 0.0039; HT29, P = 0.0003) (Fig. 4D). The combination therapy produced complete remission in about 20% of the mice in both the CT26WT and B16F10 models. The cured CT26WT-bearing mice that had received the combination regimen subsequently became immune to rechallenge with the same cancer cells (Fig. 4E). However, when cured CT26WT-bearing mice were challenged with a foreign tumor (murine 4T1 breast cancer), cured CT26WT mice grew 4T1 tumors at a rate similar to the naïve mice (Fig. 4F). Together, these results indicate the effective generation of a specific and long-lasting antitumor immunity.

Fig. 4 DMF enhances VSVΔ51 therapeutic efficacy in syngeneic and xenograft tumor models.

(A to D) CT26WT, B16F10, and HT29 tumor-bearing mice were treated intratumorally (IT) with the vehicle [dimethyl sulfoxide (DMSO)] or DMF [50 mg/kg (B16F10) or 200 mg/kg (CT26WT and HT29)] for 4 hours and subsequently injected with 1 × 108 PFU of oncolytic VSVΔ51 expressing firefly luciferase or the vehicle [phosphate-buffered saline (PBS)] intratumorally. The treatment was administered two or three times, as indicated by arrows in (C). Twenty-four hours post-infection (hpi), viral replication was monitored. (A) Representative bioluminescence images of mice. (B) Quantification of luminescence in photons/s (n = 10 to 18; horizontal lines indicate means; **P < 0.01, two-tailed t test, as compared to VSVΔ51-infected condition; dashed lines represent average background intensity). (C) Tumor volume (n = 9 to 15; data are mean ± SEM; SD values are indicated in table S1; *P < 0.05, **P < 0.01, ***P < 0.001, two-way ANOVA, comparing DMF + VSVΔ51 to DMSO alone). (D) Survival was monitored over time. Log-rank (Mantel-Cox) test indicates that the combined treatment significantly prolonged survival over VSVΔ51 alone (CT26WT: n = 10 to 13, P = 0.0008; B16F10: n = 9 to 14, P = 0.0039; HT29: n = 10 to 15, P = 0.0003). (E) Tumor volume and survival were monitored after reimplantation of CT26WT in cured and naïve mice from (D) (n = 3 to 5, mean ± SD). (F) Tumor volume was monitored after implantation of 4T1 cells in CT26WT-cured and naïve mice (n = 3, mean ± SD).

FMAEs inhibit the antiviral response

To gain further insight into the possible mechanism mediating the enhancement of OVs by DMF and other FMAEs, we performed microarray gene expression analysis on 786-0 cells 24 hours after infection with VSVΔ51 in the presence or absence of DMF, DEM, DEF, or DMM. DMF-treated cells had a similar gene expression profile to cells treated with the other FMAEs tested (fig. S4A). Upon infection with VSVΔ51, multiple antiviral genes were up-regulated as expected; however, DMF led to the inhibition of many of these (IFITIM1, MX2, GBP4, IFI27, IFNA10, and CXCL10) and up-regulated various genes, including a number of redox response genes (CYP4F11, CDK5RAP2, ANXA10, HMOX1, OSGIN1, TXNRD1, AKR1B10, AKR1B15, AKR1C1, and AKR1C2) (Fig. 5, A and B). Gene Ontology (GO) term analysis revealed that FMAE treatment of infected 786-0 cells inhibited the response to virus as well as type I interferon (IFN) signaling (Fig. 5C and fig. S4B). Consistent with repression of the type I IFN response, DMF decreased activation (phosphorylation) of both STAT1 (signal transducer and activator of transcription 1) and STAT2 24 hours after infection (Fig. 5D). Additionally, the expression of the antiviral protein IFITM1 was potently repressed by DMF, whereas the VSV viral proteins were increased (Fig. 5E). Furthermore, DMF enhanced infection of VSVΔ51, but it had no impact on infection of 786-0 cells by wild-type VSV (wtVSV) (Fig. 5E). Unlike VSVΔ51, wtVSV robustly inhibits type I IFN production (30), and therefore, we would expect the effect of DMF to be redundant in the context of wtVSV if a type I IFN response is involved in eliciting the proviral effects of DMF.

Fig. 5 FMAEs inhibit antiviral cytokine production and response to type I IFN.

(A to C) Lysates of 786-0 cells treated with FMAEs and infected with VSVΔ51 expressing GFP were collected at 24 hours after infection, and RNA or protein was extracted. (A) Scatter plot showing the expression of all genes in infected 786-0 in the presence (y axis) or absence (x axis) of DMF. Blue dots represent genes up-regulated by DMF during infection, and red dots represent genes down-regulated by DMF during infection. (B) Heat map showing the expression of the differentially expressed genes belonging to the “response to virus” GO term. (C) List of top GO terms down-regulated by FMAEs during viral infection. FDR, false discovery rate. (D) Lysates of 786-0 cells, treated with DMF (150 μM) and infected with VSVΔ51 expressing GFP or left untreated, were collected at 24 hours after infection and probed for indicated proteins by Western blot. (E) 786-0 cells were pretreated for 4 hours with DMF (150 μM) or mock-treated and infected with VSVΔ51 or wtVSV (MOI, 0.01). Corresponding viral titers were determined from supernatants 48 hours after infection (n = 3; mean ± SD). (F) 786-0 cells were treated with DMF (150 μM) or MMF (1500 μM) or mock-treated for 4 hours and infected with VSVΔ51ΔG at an MOI of 1. At 12 or 16 hours after infection, supernatants were collected and used to precondition 786-0 cells for 4 hours, and then the cells were infected with VSVΔ51 or wtVSV. Corresponding viral titers were determined 24 hours after infection from supernatants (n = 3; mean ± SD). Representative fluorescence images are shown. (G and H) 786-0 cells were treated as in (A). (G) At 16 hours after infection, supernatants were collected and assayed by ELISA for IFN-β (n = 3; mean ± SD). (H) At 36 hours after infection, supernatants were collected and assayed by ELISA for IFN-α (n = 3; mean ± SD). (I) 786-0 cells were treated with FMAEs for 6 hours and with IFN-β or IFN-α for 4 hours and subsequently infected with VSVΔ51 at an MOI of 0.1. Corresponding viral titers were determined from supernatants 48 hours after infection (n = 3; mean ± SD). (J) 786-0 cells were pretreated with 200 μM DMF for 1 or 4 hours and treated with IFN-β (250 U/ml) for 30 min or left untreated. Cell lysates were extracted and probed for pSTAT1, STAT1, and β-actin by Western blot.

To gain mechanistic insight into the effect of FMAEs on the antiviral response, we examined their ability to protect against virus challenge and IFN-mediated antiviral signaling. After treatment with FMAEs (or mock), we infected cells with a spread-deficient version of VSVΔ51 that does not encode the viral G protein (glycoprotein responsible for virus budding, host cell binding, and virus entry; VSVΔ51ΔG) to suppress the formation of de novo virions (38). Cell supernatants were collected 12 or 16 hours after infection and used to pretreat target cells before infection with VSVΔ51 or wtVSV. Our results show that although the supernatant of cells infected with VSVΔ51ΔG could protect against subsequent viral infection with VSVΔ51 or wtVSV, the addition of DMF or MMF was able to completely overcome this suppressive effect for both VSVΔ51 and wtVSV (Fig. 5F). We observed through enzyme-linked immunosorbent assay (ELISA) a decrease in the production of IFN-β and IFN-α after infection in the presence of FMAEs (Fig. 5, G and H). Furthermore, treatment of cells with DMF and other FMAEs antagonized the antiviral effects of type I IFN pretreatment on VSVΔ51 infection (Fig. 5I). Together, these data suggest that FMAEs affect antiviral signaling by repressing the production of type I IFN and downstream signaling through the JAK (Janus kinase)–STAT pathway. Consistent with this, Western blotting revealed that in cells conditioned with IFN-β at the doses that inhibit VSVΔ51 infection, DMF and DEM inhibited STAT1 phosphorylation, which is involved in transcription and response to type I IFN after viral infection (Fig. 5J and fig. S4C).

FMAEs promote infection through NF-κB inhibition independently of GSH depletion

Although our data clearly implicated an effect of multiple FMAEs on the antiviral response, the molecular chain of events leading to these effects remained unclear. DMF, DEM, DEF, and MMF share a common α,β-unsaturated carbon, which is attacked by GSH in a Michael addition reaction and is implicated in the capacity of these compounds to deplete cellular GSH and activate the antioxidant response (39). We therefore tested the impact of dimethyl succinate (DMS) (Fig. 6A), which lacks this functional moiety. We found that DMS had no impact on viral output (Fig. 6, B and C), nor did the hydrolyzed form of DMS, succinate (S). In contrast with FA, DMS, and S, all other FMAEs were able to deplete GSH (Fig. 6D). However, the proviral activity of DMF was still evident after predepletion of cellular GSH by culturing cells for 10 days in the presence of buthionine sulfoximine (BSO), an inhibitor of glutamate cysteine ligase required for the synthesis of GSH (Fig. 6E).

Fig. 6 DMF inhibits NF-κB translocation upon infection.

(A) Structures of DMS, MMS, and S. (B and C) 786-0 cells were pretreated with the indicated DMF analogs for 4 hours and subsequently infected with oncolytic VSVΔ51 expressing GFP at an MOI of 0.01. (B) At 24 hours after infection, fluorescence images were taken from the infected 786-0 cells. (C) Corresponding viral titers were determined from supernatants 48 hours after infection (n = 3; mean ± SD; ***P < 0.001, one-way ANOVA, as compared to the untreated counterpart). (D) GSH concentrations were determined in 786-0 cells after a 4-hour treatment with FMAEs (n = 4; mean ± SD; ***P < 0.001, one-way ANOVA, as compared to the untreated counterpart). (E) 786-0 cells were grown in the presence or absence of BSO (2 mM) for 7 days and pretreated with DMF (200 μM) for 4 hours or left untreated and then infected with oncolytic VSVΔ51 expressing GFP at an MOI of 0.01. Corresponding viral titers were determined from supernatants 48 hours after infection (n = 3; mean ± SD). (F) Heat map showing the expression of the differentially expressed oxidative stress genes. Expression of genes was normalized to values obtained for untreated, uninfected control. (G) HMOX1 expression quantified by quantitative polymerase chain reaction (qPCR) from 786-0 cells after a 6-hour treatment with FMAEs (n = 3; mean ± SD). (H) siNRF2 knockdown 786-0 cells were treated with DMF or untreated and infected as in (E). Corresponding viral titers were determined from supernatants 24 hours after infection. RNA was extracted, and the expression of NRF2 and IFITM1 genes was quantified by qPCR (n = 3; mean ± SD). (I and J) Cytoplasmic and nuclear protein fractions were extracted from 786-0 cells treated with DMF (200 μM) for 4 hours and (I) subsequently infected with oncolytic VSVΔ51 expressing GFP at an MOI of 1 for 8 hours or (J) treated with TNFα (30 ng/ml) for 30 min. Cell lysates were probed for multiple proteins, as indicated, by Western blot. (K) 786-0 cells were pretreated with NF-κB inhibitors targeting IKK [IKK16 (10 μM) and TPCA1 (40 μM)] for 4 hours and subsequently cotreated with DMF (150 μM) and oncolytic VSVΔ51 expressing GFP at an MOI of 0.01. Corresponding viral titers were determined from supernatants 24 hours after infection (n = 3; mean ± SD; ***P < 0.001, one-way ANOVA, as compared to the untreated counterpart).

In parallel with their impact on antiviral gene expression by microarray, FMAEs also induced robust expression of multiple genes involved in the antioxidant response (Fig. 6F). In particular, HMOX1 expression, as determined by real-time polymerase chain reaction (PCR), was consistently up-regulated well over 100-fold by FMAEs (Fig. 6G), but not by treatment with FA, DMS, or S (Fig. 6G). DMF and other FMAEs induce nuclear translocation of NRF2 via covalent modification of KEAP1, which induces antioxidant genes (40). This is consistent with our observation of the induction of HMOX1 and other NRF2 target genes by FMAEs but not FA, DMS, or S (Fig. 6, F and G). To determine whether the proviral effect of DMF is dependent on NRF2 activity, small interfering RNA (siRNA) knockout against NRF2 was performed. We found that knockdown of NRF2 did not block the capacity of DMF to enhance VSVΔ51 infection or inhibit the antiviral factor IFITM1 (Fig. 6H). Furthermore, DMF was able to enhance infection in a number of cell lines harboring KEAP1 mutations [A549 (41) and CT26WT (42)] (Fig. 1D). Together, these data suggest that the ability of FMAEs to enhance infection requires the α,β-unsaturated carbon involved in GSH depletion, but that neither GSH depletion nor NRF2 activity is a likely key mediator of this phenomenon. Given that DMF inhibits lipopolysaccharide (LPS)–stimulated cytokine production by inhibiting NF-κB nuclear entry (43, 44), we looked at nuclear and cytoplasmic fractionations of infected cells. Probing for NF-κB subunit p65 revealed that upon infection or tumor necrosis factor–α (TNFα) stimulation, DMF inhibits phosphorylation and translocation of this transcription factor, involved not only in the transcription of IFN-β but also in response to type I IFN (Fig. 6, I and J). Furthermore, DMF did not improve infection in cells treated with inhibitors of IκB kinase (IKK) degradation such as chemical compounds IKK16 and TCPA1 (Fig. 6K) (45, 46).

DISCUSSION

Here, we demonstrate that FMAEs like DMF enhance OVs such as VSVΔ51 by increasing viral spread and oncolysis in resistant cancer cells. Furthermore, we show that DMF can overcome the innate immune response of cancer cells in a manner consistent with its modulation of NF-κB activity. This results in a decrease in cytokine production and the inhibition of the type I IFN response at multiple levels, including the NF-κB and JAK-STAT pathways.

FMAEs such as DMF are electrophilic; hence, they can covalently link to essential thiol groups on various proteins, including KEAP1, which activates NRF2 (15). The activation of the antioxidative transcription factor NRF2 by DMF has been long thought to be a main mechanism in the treatment of MS; however, recent studies suggest that the anti-inflammatory activity of DMF in MS treatment occurs through alternative pathways, independent of NRF2 (43, 47, 48). Consistent with this, we show that enhancement of viral infection by DMF is not mediated by the activation of NRF2. However, NRF2 signaling may nevertheless play a role in some contexts, given a recent study implicating its involvement in sulforaphane-mediated enhancement of infection by VSVΔ51 in prostate cancer cells (49).

DMF blocks the nuclear translocation and DNA binding activity of the NF-κB transactivator subunit p65 upon stimulation with TNFα or LPS (17, 43, 44, 48). Similarly to KEAP1, DMF also covalently binds p65 cysteines, and in particular Cys38 that is essential to block its nuclear translocation and transcriptional activity (28). Consistent with this, here we show that DMF is able to block the translocation of the NF-κB transcription factor p65 upon viral infection. RNA virus infection triggers the activation of interferon regulatory factor 3 (IRF3) and NF-κB, transcription factors downstream of the viral RNA sensors such as RIG-I–like receptor (50). Both IRF3 and NF-κB are required for the expression of the antiviral cytokine IFNβ (50). However, DMF does not block the translocation of IRF3 but inhibits numerous antiviral and proinflammatory cytokines, promoting the spread of viral infection within the tumor environment. Notably, the wtVSV matrix (M) protein has the ability to inhibit NF-κB activation, whereas mutation in the M protein found in VSVΔ51 abrogates this function (51). In addition, OV infection can be improved in vitro by using compounds that inhibit NF-κB signaling through inhibition of IKK, for instance (52, 53), as we have confirmed here. However, the capacity of these compounds to enhance OV activity in vivo has not been assessed, and they are not approved for human use, in contrast with DMF.

DMF inhibits the cellular response to type I IFN, in part by decreasing IFN production via the modulation of NF-κB, in addition to the inhibition of STAT1 activation upon IFN stimulation. DMF mediates suppression of STAT1 phosphorylation in dendritic cells (DCs) (54), further supporting our findings. However, the mechanism of suppression of STAT1 by DMF remains unclear. Recent studies found that oxidative stress has a marked impact on signal transduction through the JAK-STAT pathway, impairing STAT1 phosphorylation (55, 56). This could suggest a mechanism whereby reactive oxygen species (ROS) induced by DMF indirectly blocks STAT1 phosphorylation and IFN signaling.

Despite an absence of clinical data regarding an increased risk of infections in patients treated with FAEs (14, 57), recent reports of progressive multifocal leukoencephalopathy (PML), a rare disease of the central nervous system caused by the polyoma JC virus (58), were associated with long-term treatment with DMF in several patients (5963). The susceptibility and occurrence of PML in DMF-treated patients have been widely debated (6467); however, this potential causative mechanism remains to be explored. Given our observation that DMF is able to inhibit type I IFN response upon infection, as well as inhibit DC maturation (44), this suggests that long-term treatment with DMF may increase susceptibility to IFN-sensitive viruses such as polyomaviruses (68). Further investigation to this end is warranted.

Nevertheless, Tecfidera has been safely used by more than 135,000 patients worldwide since its approval. Furthermore, Fumaderm is the most frequently prescribed drug for systemic treatment of psoriasis in Germany. DMF is emerging as a promising anticancer agent that can inhibit melanoma and colon cancer (24, 25, 69). Here, we show that DMF can also enhance the therapeutic benefit of oncolytic virotherapy in vivo. The clinical availability of DMF and related FAEs and the recent approval of the first OV for the treatment of melanoma (2, 70) provide a clear path toward clinical evaluation of this promising combination therapy.

MATERIALS AND METHODS

Study design

In our hypothesis-driven study, we assessed the therapeutic potential of the clinically approved drug DMF in combination with oncolytic virotherapy in human and murine cancer cell lines (in vitro and ex vivo), in human-derived tumor cell lines, and in human tumor specimens. The effect of the combination therapy on viral infection was assessed by plaque assay. This study was extended to syngeneic and xenograft tumor models refractory to OV therapy to analyze the effects of combination therapy in vivo. In all experiments, animals were assigned to various experimental groups at random, but experimenters were not blinded. For survival studies, sample sizes of 9 to 18 mice per group were used. Mice were euthanized when tumors had reached 1500 mm3. All outliers were included in the data analysis.

Drugs, chemicals, and cytokines

Drugs, chemicals, and cytokines and their respective suppliers and solvents used in this study are listed in table S2.

Cell lines

Cells and their respective suppliers and growth media used in this study are listed in table S3. Cells were cultured in HyQ high-glucose Dulbecco’s modified Eagle’s medium (DMEM) (HyClone) or RPMI 1640 medium (Corning) supplemented with 10% fetal calf serum (CanSera) and penicillin/streptomycin (Gibco). All cell lines were incubated at 37°C in a 5% CO2 humidified incubator. All cells were tested to ensure that they are free of mycoplasma contamination.

Human-derived cell lines

Ovarian cancer primary cultures were derived from the ascites of individuals with ovarian cancer during routine paracentesis according to Ottawa Health Science Network Research Ethics Board (OHSN-REB) protocol number 20140075-01H. These cells were maintained in complete DMEM supplemented with 10% fetal bovine serum. These cultures have been characterized and cryopreserved for use as experimental models.

Melanoma primary cultures were derived from excised surgical specimens. The surgeries were performed at the Ottawa Hospital, and specimens were taken after the receipt of patient consent according to the OHSN-REB #20120559-01. Primary cultures were maintained in RPMI 1640 medium supplemented with 10% fetal bovine serum. Primary cultures were established after scalpel-mediated homogenization of tumor specimens and filtration of the homogenate through a 70-μm nylon mesh cell strainer (Thermo Fisher Scientific). Homogenate was maintained in culture with periodically refreshed medium until sufficient cellular proliferation occurred for experimental purposes. Both primary melanoma cultures have been characterized and cryopreserved for use as experimental models.

Viruses and quantification

Rhabdoviruses. The Indiana serotype of VSV (VSVΔ51 or wild type) was used throughout this study and was propagated in Vero cells. VSVΔ51 expressing GFP or firefly luciferase are recombinant derivatives of VSVΔ51 described previously (30). All viruses were propagated on Vero cells and purified on 5 to 50% OptiPrep (Sigma) gradient, and all virus titers were quantified by the standard plaque assay on Vero cells as previously described (71). The number of infectious virus particles was expressed as PFU per milliliter.

Adenovirus. The Ad5-luciferase (adenovirus serotype 5 expressing firefly luciferase) was provided by J. Gauldie (McMaster University).

Herpes simplex virus. The HSV-1 N212 expressing GFP (72) was a gift from K. Mossman (McMaster University, Canada). HSV virus titers were quantified by the standard plaque assay on Vero cells as previously described (72).

Sindbis virus. The Sindbis virus expressing GFP was a gift from B. tenOever (Icahn School of Medicine at Mount Sinai, NY). The Sindbis virus was quantified by the standard plaque assay in Vero cells. Plaques were counted 3 days after infection.

Cell viability assay

The metabolic activity of the cells was assessed using alamarBlue (Bio-Rad) according to the manufacturer’s protocol. Fluorescence was measured at 590 nm upon excitation at 530 nm using a Fluoroskan Ascent FL (Thermo Labsystems).

Microarray and analysis

786-0 cells were plated at a density of 1 × 106 in six-well dishes and allowed to adhere overnight. The next day, cells were pretreated for 4 hours with DEM (350 μM), DEF (350 μM), DMM (300 μM), DMF (200 μM), or the vehicle. After pretreatment, the cells were infected with VSVΔ51 at an MOI of 0.01 or mock-infected. Twenty-four hours after infection, RNA was collected using an RNeasy kit (Qiagen). Biological triplicates were subsequently pooled, and RNA quality was measured using Agilent 2100 Bioanalyzer (Agilent Technologies) before hybridization to Affymetrix Human PrimeView Array (The Centre for Applied Genomics, The Hospital for Sick Children, Toronto, Canada). Microarray data were processed using Transcriptome Analysis Console 3.0 under default parameters of gene-level differential expression analysis. Fold change in gene expression was calculated for each gene in relation to uninfected, untreated control. Heat maps of normalized expression values were generated using R package pheatmap. Volcano plots of gene expression values were generated using R. GO enrichment analysis was evaluated using GOrilla (73) after correction for multiple hypothesis testing (Benjamini-Hochberg). Raw and processed microarray data have been deposited in the National Center for Biotechnology Information (NCBI) Gene Expression Omnibus database (GSE97328).

Mouse tumor models

CT26WT model. Six-week-old female BALB/c mice obtained from Charles River Laboratories were subcutaneously injected with 5 × 105 syngeneic CT26WT cells suspended in 100 μl of PBS. Eleven days after implantation, tumors were treated intratumorally once with DMF [dissolved in dimethyl sulfoxide (DMSO); 200 mg/kg] or the vehicle alone. Four hours later, tumors were intratumorally injected with 1 × 108 PFU (in 25 μl of PBS) of VSVΔ51 expressing firefly luciferase.

HT29 model. Six-week-old female BALB/c nude mice were subcutaneously injected with 5 × 106 syngeneic HT29 cells suspended in 100 μl of serum-free DMEM and 100 μl of Geltrex (Thermo Fisher Scientific). When tumors grew to about 5 mm × 5 mm (11 days after implantation), mice were treated intratumorally once with DMF (dissolved in DMSO; 200 mg/kg) or the vehicle as indicated. Four hours later, tumors were injected intratumorally with 1 × 108 PFU of VSVΔ51 expressing firefly luciferase.

B16F10 model. Six-week-old female C57BL/6 mice obtained from Charles River Laboratories were subcutaneously injected with 5 × 105 syngeneic B16F10 cells suspended in 100 μl of PBS. Eleven days after implantation, tumors were treated intratumorally once with DMF (dissolved in DMSO; 50 mg/kg) or the vehicle alone. Four hours later, tumors were intratumorally injected with 1 × 108 PFU (in 25 μl of PBS) of VSVΔ51 expressing firefly luciferase.

Tumor sizes were measured every other day using an electronic caliper. Tumor volume was calculated as (length2 × width)/2. For survival studies, mice were euthanized when tumors had reached 1500 mm3. For in vivo imaging, an IVIS (PerkinElmer) was used as described previously (35). The bioluminescent signal intensities for each mouse were quantified using Living Image v2.50.1 software. Sample size in all animal experiments was n ≥ 5. Mice were randomized to the different treatment groups according to tumor size in all experiments. Mice with no palpable tumors on initial treatment day were excluded from the study. The investigators were not blinded to allocation during experiments and outcome assessment. All experiments were performed in accordance with the University of Ottawa Animal Care and Veterinary Service guidelines for animal care under the protocols OHRI-2265 and OHRI-2264.

Ex vivo mouse model

BALB/c mice were implanted with subcutaneous CT26WT. Mice were sacrificed after tumors had reached at least 10 mm × 10 mm in size. Tumor, lung, spleen, and brain tissue were extracted from the mice, cut into 2-mm-thick slices, and cored into 2 mm × 2 mm pieces using a punch biopsy. Each tissue core was incubated in 1 ml of DMEM supplemented with 10% fetal bovine serum and 30 mM Hepes at 37°C in a 5% CO2 humidified incubator. Cores were treated for 4 hours with the indicated concentrations of chemical compounds. Subsequently, the cores were infected with VSVΔ51-GFP. GFP images were obtained for each core 24 hours after infection.

Ex vivo human samples

Tumor samples were acquired from consenting individuals during surgery, and specimens were manipulated as previously described (74). Approval was granted by the OHSN-REB for all studies requiring human tissue samples (OHSN-REB #2003109-01H and OHSN-REB #20120559-01). Patients provided their written informed consent in accordance with Declaration of Helsinki guidelines.

Immunoblotting

Cells were pelleted and lysed on ice for 30 min using 50 mM Hepes (pH 7.4), 150 mM NaCl, 10 mM EDTA, 10 mM Na4P2O7, 100 mM NaF, 2 mM Na3VO4, protease inhibitor cocktail (Roche), and 1% Triton X-100. For nuclear and cytoplasmic extracts, the NE-PER Nuclear and Cytoplasmic Extraction Kit (Thermo Fisher Scientific) was used according to the provided protocol. After protein determination by Bradford assay (Bio-Rad Protein Assay Solution), 20 μg of clarified cell lysates was electrophoresed on NuPAGE Novex 4-12% Bis-Tris precast gels (Thermo Fisher Scientific) using the XCell SureLock Mini-Cell System (Thermo Fisher Scientific) and transferred to nitrocellulose membranes (Hybond-C, Bio-Rad). Blots were blocked with 5% bovine serum albumin or milk and probed with antibodies specific for phospho-STAT1 (Tyr701, #9171, Cell Signaling Technology, used at 1:1000), STAT1 (#9172, Cell Signaling Technology, used at 1:1000), STAT2 (#72604, Cell Signaling Technology, used at 1:1000), phospho-STAT2 (#88410S, Cell Signaling Technology, used at 1:1000), IFITM1 (#60074-1-Ig, Proteintech Group, used at 1:1000, in 5% milk), VSV (a gift from E. Brown, used at 1:2000), HMOX1 (#70081, Cell Signaling Technology, used at 1:2000), or β-actin (#4970, Cell Signaling Technology, used at 1:1000). Blots were then probed with goat anti-rabbit or mouse peroxidase-conjugated antibodies (Jackson ImmunoResearch Laboratories). Bands were visualized using the SuperSignal West Pico Chemiluminescent substrate (Thermo Fisher Scientific).

Enzyme-linked immunosorbent assay

786-0 cells plated in 12-well dishes were pretreated with compound or the vehicle for 4 hours and subsequently infected with VSVΔ51-GFP at the indicated MOI or left uninfected. Cell supernatants were collected at different times after infection as indicated. IFN-α and IFN-β quantifications were performed using the VeriKine Human IFN-α or IFN-β ELISA kits (PBL Assay Science) as per the manufacturer’s instructions. Absorbance values at 450 nm were measured on a Multiskan Ascent microplate reader (MTX Lab Systems).

Quantitative real-time PCR

786-0 cells were treated for 6 hours with the indicated chemical compound or vehicle. Cells were collected and RNA extraction was performed using the RNeasy Kit (Qiagen). RNA quantity and purity were assessed using a NanoDrop ND-1000 spectrophotometer (Thermo Fisher Scientific). RNA was converted to complementary DNA (cDNA) with RevertAid H Minus First Strand cDNA Synthesis Kit (Thermo Fisher Scientific). Real-time PCRs were performed according to the manufacturer’s protocol with the QuantiTect SYBR Green PCR Kit (Qiagen) on a 7500 Fast Real-Time PCR system (Applied Biosystems). Gene expression was calculated relative to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) or β-actin. Fold induction was calculated relative to the untreated/uninfected samples for each gene. The following quantitative PCR primers were used: GAPDH, ACAGTCAGCCGCATCTTCTT (forward) and GTTAAAAGCAGCCCTGGTGA (reverse); HMOX1, ACTGCGTTCCTGCTCAACAT (forward) and GGGGCAGAATCTTGCACTTT (reverse); NRF2, CAACTACTCCCAGGTTGCCC (forward) and AGTGACTGAAACGTAGCCGA (reverse); and IFITM1, CCGTGAAGTCTAGGGACAGG (forward) and GGTAGACTGTCACAGAGCCG (reverse).

Supernatant transfer experiment

786-0 cells plated in 12-well dishes were pretreated with FMAEs or the vehicle for 4 hours and subsequently infected with VSVΔ51ΔG-GFP at an MOI of 1. This virus can infect cells and replicate its genome but does not bud or spread further because of the lack of the viral G protein, thus preventing release of viral particles in the supernatant (75). One hour after infection, the supernatant was removed to eliminate residual drug and virus and replenished with growth medium supplemented with 10% fetal bovine serum. Twelve or 16 hours after infection, supernatants were collected before being transferred to fresh 786-0 cells and processed for further analysis.

Small interfering RNA

786-0 cells plated in 12-well dishes were transfected with siRNAs (100 nM) against NRF2 (ON-TARGETplus NFE2L2 siRNA, #L-003755-00-0005, GE Dharmacon) or with a nontargeting scrambled siRNA (ON-TARGETplus Non-targeting Control Pool, #D-001810-10-05, GE Dharmacon). Transfections were carried out according to the manufacturer’s protocol (Oligofectamine, Life Technologies).

GSH assay

786-0 cells plated in a 96-well plate were pretreated with FMAEs or the vehicle for 4 hours, and the GSH concentrations were determined using the GSH-Glo Glutathione Assay Kit (Promega) as per the manufacturer’s instructions. The luminescence-based assay is based on the conversion of a luciferin derivative into luciferin in the presence of GSH, catalyzed by glutathione S-transferase. The signal generated in a coupled reaction with firefly luciferase is proportional to the amount of GSH present in the sample. The assay result was normalized using GSH standard solution provided with the kit. Luciferase expression was then measured on a Synergy Mx Microplate Reader (BioTek).

Statistics

Statistical significance was calculated using Student’s t test with Welch’s correction or one-way or two-way ANOVA test, as indicated in the figure legends. For all statistical analyses, differences were considered significant when a P value was below or equal to 0.05. The log-rank (Mantel-Cox) test was used to determine significant differences in plots for survival studies. Statistical analyses were performed using GraphPad Prism 6.0 and Microsoft Excel.

SUPPLEMENTARY MATERIALS

www.sciencetranslationalmedicine.org/cgi/content/full/10/425/eaao1613/DC1

Fig. S1. DMF promotes viral spread.

Fig. S2. FMAEs enhance VSVΔ51 infection in 786-0 and CT26WT cancer cell lines.

Fig. S3. FMAEs enhance VSVΔ51 spread in cancer cells.

Fig. S4. FMAEs inhibit response to type I IFN.

Table S1. SD values for tumor volumes in Fig. 4C.

Table S2. List of drugs, chemicals, and cytokines used in this study.

Table S3. List of cell lines used in this study.

REFERENCES AND NOTES

Acknowledgments: We thank J. Gauldie and K. Mossman (McMaster University), B. tenOever (Icahn School of Medicine at Mount Sinai), E. Brown (University of Ottawa), and M. Holcik (Children’s Hospital of Eastern Ontario Research Institute) for providing Ad5-luciferase, HSV viruses, Sindbis virus, VSV antibody, and the 76-9 cell line, respectively. We also thank N. Forbes and R. Arulanandam for reviewing the manuscript and C. Boddy (University of Ottawa), T. Alain (Children’s Hospital of Eastern Ontario Research Institute), D. Gray [Ottawa Hospital Research Institute (OHRI)], and M.-E. Harper (University of Ottawa) for their helpful advice. Funding: M.S. is supported by a Canadian Institute for Health Research (CIHR) doctoral fellowship. J.-S.D. is supported by the CIHR New Investigator Award—Infection and Immunity (INI-147824). J.-S.D. and J.C.B. hold grants from the Terry Fox Research Institute (TFF 122868) and the Canadian Cancer Society supported by the Lotte & John Hecht Memorial Foundation (703014). R.K. was supported by an Ontario Graduate Scholarship. A.B. was supported by a BioCanRx summer studentship award. Author contributions: M.S., P.O., C.R., A.B., R.K., and L.P. conducted in vitro experiments. M.S. and A.C. performed mouse experiments. M.S. and B.A.K. processed patient tumor biopsies. B.A.K. and C.I. isolated human-derived cell lines. M.S. and J.-S.D. participated in the conception and design of the study. M.S. and J.-S.D. wrote the manuscript with editorial contributions from L.P., R.K., and B.A.K. J.-S.D. and J.C.B. supervised the study. Competing interests: J.C.B. is a founder and on the board of directors of Turnstone Biologics. M.S., R.K., and J.-S.D. are inventors on the patent application submitted by OHRI that covers the use of FMAEs in combination with OVs (U.S. Provisional Patent Application No. 62/590,456, filed 24 November 2017). Data and materials availability: Microarray data have been deposited in the NCBI Gene Expression Omnibus database (GSE97328).
View Abstract

Navigate This Article