Clonal Variation in Interferon Response Determines the Outcome of Oncolytic Virotherapy in Mouse CT26 Colon Carcinoma Model

ORIGINAL ARTICLE Clonal variation in interferon response determines the outcome of oncolytic virotherapy in mouse CT26 colon carcinoma model

JJ Ruotsalainen1, MU Kaikkonen1, M Niittykoski1, MW Martikainen1, CG Lemay2, J Cox2, NS De Silva2, A Kus2, TJ Falls2, J-S Diallo2, F Le Boeuf2, JC Bell2, S Ylä-Herttuala1, AE Hinkkanen1 and MJ Vähä-Koskela2

In our earlier studies, Semliki Forest virus vector VA7 completely eliminated type I interferon (IFN-I)-unresponsive human U87-luc glioma xenografts, whereas interferon-responsive mouse gliomas proved refractory. Here, we describe in two clones of CT26 murine colon carcinoma, opposed patterns of IFN-I responsiveness and sensitivity to VA7. Both CT26WT and CT26LacZ clones secreted biologically active interferon in vitro upon virus infection but only CT26WT cells were protected. Focal infection of CT26WT cultures was self-limiting but could be rescued using IFN-I pathway inhibitor Ruxolitinib or antibody against IFNβ. Whole transcriptome sequencing (RNA-Seq) and protein expression analysis revealed that CT26WT cells constitutively expressed 56 different genes associated with pattern recognition and IFN-I signaling pathways, spanning two reported anti-RNA virus gene signatures and 22 genes with reported anti-alphaviral activity. Whereas CT26WT tumors were strictly virus-resistant in vivo, infection of CT26LacZ tumors resulted in complete tumor eradication in both immunocompetent and severe combined immune deficient mice. In double-flank transplantation experiments, CT26WT tumors grew despite successful eradication of CT26LacZ tumors from the contralateral flank. Tumor growth progressed uninhibited also when CT26LacZ inoculums contained only a small fraction of CT26WT cells, demonstrating dominance of IFN responsiveness when heterogeneous tumors are targeted with interferon-sensitive oncolytic viruses.

Gene Therapy (2015) 22, 65–75; doi:10.1038/gt.2014.83; published online 18 September 2014

INTRODUCTION fluorescence marker EGFP to probe defined mixtures of CT26WT Despite the recent advances in modern cancer therapies, and CT26LacZ cells which, as we report here, represent a pair of prognoses of advanced cancers remain poor. Oncolytic virother- IFN-I-responsive and -unresponsive tumor clones, respectively. apy presents an attractive option to chemo-, radio- and Resistance of CT26WT cells to virus spread in vitro was dependent monoclonal antibody therapies, as oncolytic viruses display low on IFN-I and could be overcome by neutralization of secreted β toxicity and may induce strong anti-tumor immune responses.1,2 interferon beta (IFN ) or by interference with intracellular IFN fi Most viruses are sensitive to type I interferon (IFN-I) and many signaling. Oncolytic ef cacy in intradermal CT26WT and CT26LacZ oncolytic viruses replicate more efficiently in tumors displaying tumors in immunocompetent mice was 0 and 100 percent, constitutive Ras-pathway activation, which facilitates oncolytic respectively. Furthermore, we describe here, to our knowledge the fi virus replication by interfering with virus-mediated IFN-I rst time in the context of oncolytic virotherapy, a constitutive induction.3,4 Genes involved in IFN-I signaling are often tran- global antiviral gene signature comprising besides two documen- scriptionally repressed in tumors as they play a role in growth ted anti-RNA virus signatures, the OAS/Mx and immune-related restriction, apoptosis and immunogenicity, and one of the GTPases (IRGs), also several other known anti-alphavirus effectors and IFN-I signal mediators. principal predictive markers for oncolytic virus efficacy is the limited capacity of the tumor to mount antiviral defense in response to IFN-I.5–11 However, although heterogeneity in IFN-I signaling among cell lines and patients is apparent, the RESULTS implications for virotherapy of patient-specific intratumoral Balb/c CT26 colon carcinoma clones differ in their responsiveness heterogeneity have not been studied. It has been unclear whether to IFN-I and susceptibility to infection clonal selection of resistant subpopulations, known to confer Spurred by the contrasting sensitivity of different glioma cells to resistance to other advanced therapies,12 may also hamper oncolytic VA7,17,18 we screened other cancer cell lines in our oncolytic virotherapy, and preclinical models for such phenom- laboratories as part of an effort to identify variables affecting enon have not been described. permissiveness to oncolytic viruses. We discovered that two To address these shortcomings, we have used IFN-I-sensitive clones of CT26 mouse (Balb/c) colon carcinoma cells originally oncolytic Semliki Forest virus (SFV) VA711,13–17 encoding generated in Nicholas Restifo’s laboratory,19 differed in sensitivity

1A.I. Virtanen Institute for Molecular Sciences, Department of Biotechnology and Molecular Medicine, University of Eastern Finland, Kuopio, Finland and 2Centre for Innovative Cancer Therapeutics, Ottawa Hospital Research Institute, Ottawa, Canada. Correspondence: Professor AE Hinkkanen, A.I. Virtanen Institute for Molecular Sciences, University of Eastern Finland, P.O. Box 1627, Neulaniementie 2, Kuopio 70211, Finland. E-mail: ari.hinkkanen@uef.fi Received 7 March 2014; revised 11 June 2014; accepted 6 August 2014; published online 18 September 2014 Clonal dominance in oncolytic virotherapy JJ Ruotsalainen et al 66 to replication and oncolysis by VA7 both in vitro and in vivo, with equivalents) ISG15, ISG20, Rsad2 (Viperin), Gbp2, Irgm2 (GTPI), the retrovirally transduced CT26LacZ cells displaying greater Igtp, Tgtp1/2, Ifi47 (IRG47), Irgm1 (LRG-47), Iigp1, Rig-I, PKR, IRF7, susceptibility to the virus than the non-transduced CT26WT Ifitm3, MDA5, Gm6907 (Phf11), IRF9, Gbp4, Ifi44l, Ube2l6 and OAS3 counterparts. Specifically, VA7-EGFP replicated in (Figure 1a, have been reported21–25 to have direct anti-alphaviral activity Supplementary Table S1) and killed (Figure 1b) CT26LacZ cells alone or in combination with zinc antiviral protein that was more effectively than CT26WT cells, and intratumoral virus expressed at equal levels in both of the cell lines (Figure 3a, injection resulted in complete eradication of CT26LacZ tumors Table 1). The genes expressed at higher levels in CT26WT cells but had no impact on CT26WT tumor growth (Figure 1c). As our compared with CT26LacZ cells included the central IFN-I signaling previous work with murine gliomas strongly implicated IFN-I- mediators STAT1 and IRF9, and in addition five IRGs constituting a mediated restriction of virus infection, we tested VA7 infectivity recently reported anti-alphaviral gene signature.26 Furthermore, and killing capacity in the presence and absence of IFNβ. we observed constitutive expression of OAS and Mx ISGs, which Experiments revealed robust infection of both cell lines if left have been proposed as a resistance gene signature against untreated, but upon IFNβ pretreatment almost complete inhibi- another IFN-I sensitive virus, VSV.27 tion of virus replication and oncolysis only in the CT26WT cells Interestingly, VA7 infection did not cause statistically significant (Figure 1d). Infectivity of both VA7 and another IFN-I-sensitive transcriptomic changes in CT26WT cells (Figure 3a), not even in virus, vesicular stomatitis virus (VSV), was inhibited by IFNβ in a IFNβ mRNA levels despite that these cells released IFNβ protein dose-dependent manner in CT26WT cells whereas CT26LacZ cells upon prolonged infection (Figure 2b). This might be explained by failed to mount complete antiviral defense even at 10 000 U ml − 1 the early sampling time point (12 h p.i.) we chose to avoid the IFNβ (Supplementary Figure S1). transcriptional shutdown and lysis of sensitive CT26LacZ cells and/ or by the constitutive expression of the antiviral factors listed Both CT26 clones secrete IFNβ but only CT26WT cells are capable above which may reduce the need for de novo transcriptional of effectively transmitting IFNβ-dependent antiviral signaling activation. To further characterize the IFN-I sensitivities of the two clones and to mimic in vivo situation, we used a single-focus expansion Impaired antiviral defense in CT26LacZ cells is associated with low assay20 that restricts virus replication to radial growth beneath an STAT1 levels and may be restored by non-specific transfection agarose overlay in the presence or absence of neutralizing To confirm our transcriptomic findings, we probed by western (polyclonal) antibodies to IFNα or IFNβ. Results revealed that blotting select differentially transcribed IFN-I signaling mediators although VA7 is capable of infecting cells at the inoculation site, it in CT26WT and CT26LacZ cells at multiple time points p.i. or post is nevertheless incapable of spreading radially in CT26WT cells recombinant mIFNβ exposure. As an immediate downstream because of paracrine effects of IFNβ (Figure 2a). Furthermore, we indicator of IFN-I receptor activation we chose STAT1, which found that though CT26LacZ cells did not effectively respond to allows for detection of phosphorylation-dependent activation, and IFN-I, they were able to produce it upon VA7 infection (Figure 2b). as an IFN-I-inducible effector we chose OAS, known to be This interferon was biologically active, as cell culture supernatants upregulated in virus-resistant GL261 gliomas upon VA7 from infected CT26LacZ cells protected CT26WT cells against VA7 infection17 and previously identified as a tumor resistance marker infection (Figure 2c). Furthermore, whereas all CT26LacZ cells were against oncolytic VSV.27 Our results revealed a rapid and transient infected at low multiplicity of infection (MOI) (10 plaque-forming increase in STAT1 phosphorylation in CT26WT cells 3 h after units (PFUs) per 50 000 cells) by 48 h post infection (p.i.), the same mIFNβ addition compared with no changes observed in CT26LacZ virus dose in CT26WT cells resulted in a self-limiting infection that cells (Figure 3b), confirming proper IFN-I signal transduction in the could only be rescued using JAK/STAT pathway inhibitor former. We also observed a very high constitutive synthesis of Ruxolitinib (Figure 2d, Supplementary Figure S2). Interestingly, both STAT1 and OAS in CT26WT cells compared with CT26LacZ Ruxolitinib enhanced the infection even further when the cells cells, corroborating the transcriptomic data and canonical path- received small dose IFNβ pretreatment, an unanticipated syner- way assessments (Table 1), and pointing to a lowered antiviral getic effect we are currently investigating. Notably, IFNβ could defense in the CT26LacZ clone. also somewhat protect CT26LacZ cells when only miniscule 10 To study whether the lower STAT1 protein content in the PFU amount of virus was used suggesting that the IFN response CT26LacZ cells compared with CT26WT cells (Figure 3) was defects in CT26LacZ cells are quantitative, not qualitative in responsible for the dysfunctional antiviral response, we reintro- nature. duced STAT1 expression into CT26LacZ cells by plasmid transfection and performed focal infection with a miniscule 10 PFU virus Differential capacity to support oncolytic virus replication dose. While the virus resistance phenotype was indeed restored correlates with unique global gene expression signatures upon STAT1 plasmid transfection, surprisingly also control plasmid transfection increased OAS1 and STAT1 gene expression up to the To gain a deeper understanding of the differences in permissive- level of CT26WT cells and conferred protection against the focal ness and antiviral signaling between the two CT26 cell clones, we infection (Supplementary Figure S4). carried out comparative RNA-Seq whole transcriptome analysis of infected and uninfected CT26WT and CT26LacZ cells. Both cell lines expressed the colon carcinoma associated markers Ict1, VA7 destroys CT26LacZ tumors in vivo via focal infection and Sdccag8, Sdccag3, Marcksl1, Lamp2, plasminogen activator causes bystander necrosis independent of the adaptive immune (tissue) and caspase 6 at equal levels. Furthermore, CT26WT and system CT26LacZ cells displayed high correlation in their gene expression To gain understanding of the biological significance of the (R = 0.98 as opposed to R = 0.64 when CT26WT is compared with differential antiviral capacities of the two CT26 tumor clones, we unrelated cells), which was not surprising, as the two clones share characterized immunohistochemically tumor samples from virus- common origin (Supplementary Figure S3). Results showed that a treated and uninfected control animals. We observed a perivas- total of 1622 mRNAs constituting approximately 10% of all the cular pattern of infection in the CT26LacZ tumors with extensive sequenced messengers were differentially transcribed between necrotic areas surrounding the sites of infection (Figure 4a). uninfected CT26WT and CT26LacZ cells. Compared with CT26LacZ A similar infection pattern was previously observed for oncolytic cells, CT26WT cells displayed higher constitutive expression of 56 VSV.28 In contrast, we saw no evidence of CT26WT tumor infection different genes of inflammation, pattern recognition receptor and at any time point p.i. (Figure 4a). These results were corroborated IFN-I signaling pathways. Twenty-two of them (or their human ex vivo, as live tissue slices only from CT26LacZ tumors could be

Figure 1. VA7-EGFP infects and kills CT26LacZ tumors better than CT26WT tumors in vitro and in vivo correlating with responsiveness to exogenous IFNβ.(a) Virus production (PFU ml − 1 in cell culture supernatant) from CT26LacZ cells infected at MOI = 0.01 is faster than from CT26WT cells. Graph shows absolute titers from one representative experiment out of two. (b) VA7 kills CT26LacZ cells more effectively than CT26WT cells as evidenced by Alamar blue cell viability assay 48 h p.i., *Po0.05 (unpaired, two-tailed t-test). (c) A single intratumoral injection of either phosphate-buffered saline or 1 × 108 PFU VA7 virus in palpable intradermal tumors in groups of Balb/c mice (N = 5) revealed opposing sensitivity of CT26LacZ and CT26WT tumors to VA7. (d) VA7-EGFP (MOI = 1 and MOI = 0.01) infects and kills both CT26WT and CT26LacZ cells in 48 h p.i. as shown by ﬂuorescence and phase contrast micrographs, respectively. The cytotoxic effect is more complete in CT26LacZ cells with both MOIs used, and exogenous IFNβ pre-treatment (1000 U ml − 1) protects efﬁciently only CT26WT cells. Scale bar: 200 μm.

Figure 2. Although both CT26WT and CT26LacZ cells produce IFNβ in infection, it restricts the VA7 replication only in CT26WT cells. (a) Plaque expansion assay on CT26LacZ and CT26WT cells showed that radial spread of VA7 under spatially restricted conditions (agarose overlay) from a central inoculation spot occurred in CT26WT only when virus was co-administered with polyclonal neutralizing anti-IFNβ antibody, demonstrating IFNβ-dependent limiting VA7 infection of CT26WT cells. Conversely, CT26LacZ cells supported progressive infection and spread by VA7. (b) Both CT26WT and CT26LacZ cells efficiently produce mIFNβ as shown by ELISA assay for CT26WT and CT26LacZ cell supernatants after VA7-EGFP infection. The used MOI was 0.1 and supernatants were collected 48 h p.i. Each bar marks the average of four replicas per cell type (two infection replicas and two technical ELISA replicas) and the error bars their s.d. The difference is statistically significant (Po0.0001) as analyzed by unpaired, two-tailed t-test. mIFNβ levels in supernatants obtained from uninfected CT26WT and CT26LacZ cells were below the ELISA detection limit 31.2 pg ml − 1 (not shown). (c) Pretreatment of CT26LacZ or CT26WT cells with homologous or heterologous virus-free (filtered) conditioned medium collected from VA7 infected cells 48 h p.i. (MOI = 0.01) revealed secretion of interferon into the supernatant by both cell types. However, only CT26WT cells had the capacity to mount antiviral defense, as assessed by infecting conditioned-medium-pretreated cells at MOI = 0.1, × 10 magnification. (d) Miniscule 10 PFU amount of VA7-EGFP results in productive infection only in CT26LacZ cells whereas the self-limiting infection can be rescued with JAK/STAT inhibitor Ruxolitinib in CT26WT cells counteracting also the effects of exogenous IFNβ (100 U ml − 1) pre-treatment (the full micrograph data are available in Supplementary Figures 2). Scale bar: 200 μm.

Figure 3. CT26WT cells display higher basal expression of several antiviral genes than CT26LacZ cells at transcript and protein levels. (a) RNA- Seq heatmap of differentially regulated genes (centered to mean expression) of CT26WT and CT26LacZ infected (INF) and control (Ctrl) cells on pattern recognition receptor, IFN and inflammation pathways. Select genes without statistically significant (Po0.01) changes were included in the heatmap as they are important mediators of IFN-I signal and are marked (1). The 22 genes (or their human equivalents) that have been previously reported to have direct anti-alphavirus activity are marked with (2) (as Balb/c Mx2 is inactivated, Mx1 is not marked, whereas human MxA anti-SFV activity has been reported50), five IRG genes that constitute the IFNβ inducible RNA-virus defence gene signature are marked (3). Genes marked (4) are pseudogenes. Each sample was prepared and sequenced as two replicates. Cutoffs: RPKM40.5, Po0.01. (b) OAS or total STAT1 levels are not induced upon VA7 infection (MOI = 0.1) in CT26WT cells. In contrast, the high basal expression level seems to decline as a result IFN beta (100 U ml − 1) induced STAT1 phosphorylation and recovers back to basal level at the later time points. OAS, total STAT1 and STAT1-P remained low in CT26LacZ cells regardless of IFNβ treatment or VA7 infection and this was true also for CT26.CL25 cells (fresh CT26LacZ batch from ATCC).

© 2015 Macmillan Publishers Limited Gene Therapy (2015) 65 – 75 Clonal dominance in oncolytic virotherapy JJ Ruotsalainen et al 70 between the two CT26 clones exist that can be exploited to Table 1. Ingenuity IPA pathway analysis conﬁrms that CT26WT and generate protective immunity.29,30 CT26LacZ cells differ in their antiviral signaling

Name P-value Ratio Small fraction of virus-resistant cells determines treatment outcome in clonally mixed CT26 tumors Top canonical pathways fi Activation of IRFa by cytosolic pattern 2,38E-03 10/42 Finally, to assess the impact on therapy ef cacy of a hetero- recognition receptors geneous disease composed of sensitive and resistant cancer cell Interferon signaling 3,14E-03 7/24 clones, we mixed CT26LacZ and CT26WT cells at varying ratios and Role of JAK1b and JAK3 in γcc cytokine signaling 4,08E-03 10/45 infected them in vitro and in vivo. Cell infection rate and death TR/RXRd activation 4,35E-03 12/60 correlated linearly with the increasing proportion of sensitive LXR/RXRe activation 5,01E-03 12/62 CT26LacZ cells in vitro (Figure 6a). However, and as earlier 31 Top 2 canonical pathways found in Ingenuity IPA core analysis when reported with chemotherapy, the proportion of therapy- comparing RNA-Seq data between uninfected CT26WT (n = 2) and sensitive clones did not correlate linearly with survival in vivo, CT26LacZ (n = 2) cells were IRF activation by cytosolic pattern recognition and presence of even a small fraction of resistant CT26WT cells receptors and interferon signaling. aInterferon regulatory factor. bJanus resulted in cancer progression in the majority of mice (Figure 6b). kinase. cGamma chain. dThyroid receptor/Retinoid X receptor. eliver X receptor/Retinoid X receptor. DISCUSSION fi Here, we have demonstrated dramatic difference in the antiviral ef ciently infected (Figure 4b). Finally, to assess the role of the signaling of two mouse colon carcinoma clones CT26WT and adaptive immune system in destruction of CT26LacZ tumors, we CT26LacZ having the same origin, and highlighted the implica- compared oncolytic potency of VA7 in severe combined immune tions of these phenotypic differences for oncolytic virotherapy fi de cient (SCID) mice devoid of T- and B-cell immunity with in vitro, in vivo and ex vivo. Several studies have shown that standard immunocompetent Balb/c mice. Results showed that established cancer cell lines differ from each other in their IFN-I irrespective of the adaptive immune status of the host, all VA7- responsiveness and permissiveness to oncolytic viruses.5–11,32 treated CT26LacZ tumors regressed completely (Figure 4c). Thus, Intratumoral heterogeneity in IFN-I signaling components, such as the therapeutic efficacy was independent of adaptive immunity, IFN-I receptor and STAT1, has also been observed upon 11 which is in line with our earlier findings in U87-luc xenograft and immunohistochemical analysis of human tumors.33–36 Further- 17 syngeneic GL261 models. more, A431 human epidermoid carcinoma cell line was found to be clonally heterogenic in STAT1 expression.37 However, the CT26LacZ and CT26WT cells differ in immunogenicity and display implications of this heterogeneity for oncolytic virotherapy only partial immune overlap have remained elusive, as the substantial number of IFN- fi All those Balb/c mice cured of CT26LacZ tumors were protected defective cancer cell lines identi ed in vitro does not seem to fl against rechallenge with homologous cells 2 months later, but not re ect the minimal portion of patients who display viral oncolysis against heterologous challenge by Balb/c mouse 4T1 breast in the clinics. One explanation for this discrepancy could be the carcinoma cells (not shown). Furthermore, we observed occasional inability of dissociated cell cultures to reveal low-level paracrine spontaneous tumor regression in the CT26LacZ, but not in the antiviral IFN-I signaling. Indeed, although dissociated CT26WT cells were readily infectable (Figure 1d), densely packed cells under CT26WT mock-treated (saline) mice only when the recipient mice β were immunocompetent, suggesting that the LacZ expressing agarose quickly stopped the virus spread (Figure 2a). IFN tumors were more immunogenic than CT26WT tumors (Figure 4c). pretreatment suppressed besides VA7-EGFP infection also the infection of another type I IFN-sensitive RNA virus VSV-D51-EGFP To confirm the observations, we administered to groups of mice in CT26WT but not in CT26LacZ cells (Figure 1d, Supplementary (n = 5) intraperitoneally either phosphate-buffered saline or Figure S1). Moreover, whereas live tissue slices prepared out of 30 Gy-irradiated CT26LacZ or CT26WT cells (5 × 106 cell equivalents) CT26LacZ tumors were readily infectable, the respective CT26WT on days 0 and 7. Two weeks later, mice received an intradermal slices were virus-resistant suggesting that tumor stroma probably injection of 3 × 105 live tumor cells (the same cell clone they were contributed to virus resistance in vivo (Figure 4b). In line with our immunized with) and were followed for up to 2 months. findings, hepatomas having defects only in IFN-I induction were Indeed, CT26LacZ cells afforded protection to 100% of mice reported to be resistant against the Sindbis virus, a close relative against homologous challenge, whereas immunization with of SFV, likely because of alternative interferon sources in the irradiated CT26WT cells only protected 40% (2/5 mice) tumor microenvironment, whereas tumors with impaired IFN-I (Supplementary Table S2). response due to receptor knockout were highly sensitive.38 To further investigate whether the two CT26 clones retain any Some cell lines such as NIH 3T3 fibroblasts have been reported cross-protective antigens and to probe the kinetics of elicited to contain a heterogenic population of interferon-responsive and protective immune responses upon viral oncolysis, we challenged -unresponsive cell clones,39 and this might be the case also in virus-treated CT26LacZ tumor-bearing animals at different time primary human tumors.33–35 We have shown here that this points with CT26WT cells. We implanted into the right flank of heterogeneity quickly results in virus-resistance and tumor escape 5 5 mice 2 × 10 CT26LacZ cells and the left flank with 2 × 10 (Figure 6). Tumor-promoting and antiviral signaling pathways CT26WT cells, the latter either at the same time, 8 days or occasionally overlap. It was shown that STAT1 can confer 2 months after CT26LacZ implantation. Results showed that resistance to genotoxic stress, such as radiation and chemother- CT26WT cells grew unperturbed irrespective of whether they were apy, and may stimulate cancer cell proliferation.40–42 Thus, it is implanted at the same time (not shown) or 8 days after CT26LacZ possible that STAT1 activation in such tumors would also induce cell implantation and VA7 treatment (Figures 5a and b). However, defenses against oncolytic viruses rendering them refractory to when CT26WT cells were implanted approximately 2 months after the therapy, although this remains to be shown. CT26LacZ tumors had been completely eliminated by VA7 virus, Virus resistance of CT26WT cells was associated with constitutive two out of four mice rejected the challenge (Figure 5c), suggesting transcription of numerous genes related to inflammation, pattern that although LacZ has been reported to be the major recognition receptor and IFN signaling pathways, including known immunodominant antigen in the CT26LacZ cells, shared antigens antiviral OAS/MxA and IRG gene signatures, and two central

Figure 4. Infection of CT26WT tumors is inhibited when tumor stroma is present whereas CT26LacZ tumors are readily infected and eliminated regardless of adaptive immunity. (a) VA7-EGFP intratumoral injections (1 × 106 PFU) lead to progressive infection (anti-SFV primary Ab, brown color) and destruction of the CT26LacZ tumors of Balb/c mice, whereas no virus was found in any of the 24, 48 or 96 h p.i. time points in CT26WT tumors. The initial site of CT26LacZ tumor infection seemed to be the tumor tissue around the injection needle cavity, whereas perivascular infection pattern (arrows) accompanied by appearance of massive necrotic (N) centers is seen especially at the late 96-h time point. Tumor collected 24 h post saline injection was used as a control. Scalebar: 2000 μm. (b) Live slices obtained from CT26LacZ tumors but not from CT26WT tumors were infected efficiently within 40 h of VA7-EGFP application (107 PFU). The pictures were taken using multiphoton × 25 objective and are representative of four infection replicates. (c) Intratumoral VA7-EGFP injection destroyed rapidly the CT26LacZ tumors both in immunocompetent and in immunocompromised mice. Virus (106 PFU in 10 μl volume) or 10 μl of saline (mock) was injected when tumors were palpable at timepoint 0. Virus-treated Balb/c (P = 0.0103*, Log-rank (Mantel-Cox) test) and SCID mice (Po0.0001***) survived significantly longer than mock treatment groups. mediators of IFN-I response, STAT1 and IRF9 (Figures 3a and b and to induction of its expression as a positive feedback mechanism, Table 1).26,27 This is to our knowledge the first report of a global and also the unphosphorylated STAT1 has been reported to drive antiviral gene signature in context of oncolytic virotherapy. long-term expression of certain ISGs.43 Intriguingly, the clear Interferon response leads besides to STAT1 phosphorylation also majority of the ISGs that were reported to be induced directly by

Figure 5. VA7-mediated oncolysis of CT26LacZ tumors fails to prevent growth of CT26WT tumors implanted in the same animal. (a)VA7 injection in palpable intradermal CT26LacZ tumors implanted on the right flank of animals resulted in complete eradication of these tumors in all animals, but did not affect outgrowth of CT26WT tumors implanted in the same mice on the left flank either at the same time as the CT26LacZ tumors (not shown) or 8 days after CT26LacZ tumors had been implanted. (b) Control mice (to those in panel a) harboring CT26WT tumors on the left flank and injected with VA7 virus when palpable. (c) VA7 oncolysis of CT26LacZ tumors affords protective immunity against heterologous CT26WT challenge in two of four mice, challenged ~ 2 months after CT26LacZ tumors had been eradicated. Each graph in panels a, b and c represents individual tumor development. Solid lines: CT26LacZ tumors, dotted lines: CT26WT tumors.

unphosphorylated STAT1 were also constitutively expressed in response. The observed plasticity CT26LacZ cells have in their CT26WT cells including, OAS1, OAS2, OAS3, STAT1, IFI44, IFI44L, antiviral gene expression and phenotype suggests that the IFIH1 (MDA5), IFI35, IFIT3, Mx1, IRF7 and IFIT1 (p56), suggesting that differences CT26WT and CT26LacZ cells have in this regard might the high basal level of unphosphorylated STAT1 seen in the be rather gene regulatory than mutational in nature. In line with CT26WT cells might explain their constitutively primed antiviral this hypothesis, IFNβ pretreatment could inhibit VA7-EGFP state (Figure 3b). infection partially in CT26LacZ cells when a miniscule 10 PFU Despite the observed productive VA7 infection, manifested as virus amount was used (Supplementary Figure S2). This demon- viral RNA in the CT26LacZ cells used for RNA-Seq and viral strated that these cells have diminished but not abolished particles in the corresponding supernatants, (Supplementary Table response to type I IFNs and hence the difference CT26WT and S1) we did not observe induction of late type I IFNs (mainly IFN-α CT26LacZ clones display in their antiviral signaling is more species) (Figure 3a). This observation together with the fact that quantitative than qualitative in nature. inflammatory cytokine IL6 and the early IFN-I IFNβ were both We have focused in the present report on the effects of clonal induced in CT26LacZ cells upon infection (Figures 2b and 3a), disparity in tumor antiviral defenses on oncolytic virotherapy and strongly indicated that CT26LacZ tumors are compromised not in aim in the future to probe the mechanisms selecting for variable pattern recognition or IFN-secretion, but only in capacity to STAT1 expression (or other components of the antiviral defenses) respond to IFN-I (Figure 1). We postulate that this defect derives in the two CT26 clones under normal non-infected conditions. The from insufficient STAT1 levels in the CT26LacZ cells, but as the retroviral integration sites or copy number in the CT26LacZ cell antiviral effect was induced also by the control plasmid, as line are not known, and even though none of the antiviral gene documented also by others,44 we were unable to answer the transcripts presented in the heatmap was truncated according to question specifically (Supplementary Figure S4). According to our UCSC genome browser analysis (Figure 3), we cannot exclude the deep sequencing results, both CT26WT and CT26LacZ cells possibility that the retroviral integration(s) in the CT26LacZ cell express equal amounts of cytosolic DNA sensing receptors IFI16 line could have influenced the transcription profile in a more and cGAS, each of which could be an important initiator of the indirect way. On the other hand, we do not know at this stage if

Figure 6. Mixed heterogeneous CT26 model reveals dominance of virus-resistant phenotype. (a) Keeping total cell number constant, infection of mixed cultures of CT26LacZ and CT26WT cells revealed a near-linear relationship of the degree of infection and lysis to the fraction of virus susceptible CT26LacZ cells, estimated from fluorescence micrographs and fixed cell staining 48 h p.i. at MOI = 0.01, × 10 magnification. (b)In contrast to results in vitro (panel a), tumors established by implanting mixtures of CT26LacZ and CT26WT cells (2 × 105 total cells per mouse) showed dominance of virus-resistant phenotype even when CT26WT cells in the tumor inoculum constituted a small fraction of 10%. the constitutive antiviral gene signature in CT26WT cells repre- therapies against IFN-I-sensitive and virus-resistant cell popula- sents a steady-state or is the result of chronic stimulation by tions in heterogeneous cancer. unknown mechanisms. Importantly, however, as shown by Ruxolitinib and INFβ neutralization treatments of CT26WT cells and plasmid transfection of CT26LacZ cells, the antiviral defense MATERIALS AND METHODS can be manipulated either way, offering prospects for generation Cell cultures and viruses of specific molecular means to locally lower resistance to VERO(B), BHK-21 and CT26LacZ cell lines have been originally acquired virotherapy or to increase resistance against pathogenic viruses. from ATCC and maintained in the groups of John Bell and Ari Hinkkanen. We have established models of intratumoral and intertumoral CT26WT and CT26.CL25 cells were freshly acquired from ATCC. CT26LacZ heterogeneity of type I IFN response, and utilized them to and CT26.CL25 represent the same cell line, but we use here distinct demonstrate clonal antiviral dominance and the immediate nomenclature to make the separation between the passaged (CT26LacZ) and freshly acquired (CT26.CL25) batches. The cell lines were tested to be consequences it poses for oncolytic virotherapy. When intratu- mycoplasma-free. Cloning and production of VA7-EGFP has been moral heterogeneity was modeled in vivo by mixing of CT26WT described.17,46 Virus stocks and cell culture supernatants were titered by and CT26LacZ cells before transplantation, a small fraction of plaque assay on VERO(B) cells (see Supplementary Materials and Methods). virus-resistant CT26WT cells was enough to result in tumor escape fi in the majority of the animals (Figure 6). This nding thus Enzyme-linked immunosorbent assay recapitulated the well-characterized clonal resistance phenom- IFNβ secretion from cells was detected by infecting cells in 12-well plates enon observed in cancer patients receiving other advanced 12 at MOI = 0.1 and harvesting supernatants from duplicate wells 48 h p.i. and therapies. In the model of intertumoral heterogeneity, eradica- measuring by ELISA (two technical replicas per sample) according to tion of monotypic CT26LacZ tumors resulted in the induction of manufacturer’s instructions (Verikine IFNβ ELISA kit, PBL InterferonSource, cross-protective immune responses against CT26WT cells in half of 42400-1, Piscataway, NJ, USA). the treated mice. The result was more promising than was 30 achieved by Mroz et al. with photodynamic therapy in the same Plaque expansion assay setting, suggesting that targeting virus-sensitive tumors may The protocol was adapted from Duca et al.20 In short, monolayers of near- result in bystander effects against the resistant ones. The elicited confluent cells in 6-well plates were covered with 1% agarose overlay cross-protective immune responses were nevertheless incomplete supplemented with 2 × Penicillin/Streptomycin (Sigma-Aldrich, P0781, (Figure 5), possibly because of the immunodominance of LacZ St Louis, MO, USA). A ~ 3 mm diameter cylindrical well was created by antigen reported by Mroz and colleagues. The immuno- stamping out agarose with a cut 200 μl pipette tip. Virus and/or virotherapeutic approach needs thus to be modified to reach recombinant antibodies (PBL InterferonSource, Rabbit polyclonal pan-IFNα β higher level of protection with detailed characterization of the antibody, 32100-1, Rabbit polyclonal IFN antibody, 32401-1) in standard specificity and kinetics of the induced responses. Such approaches medium were carefully pipetted into the well with total volume not exceeding 10 μl. Plates were observed under fluorescence stereomicroscope could include ectopic expression of immunostimulatory cytokines (Leica M165 C, Wetzlar, Germany) with lids off. such as IL-12, a strategy that has been shown to greatly increase the SFV-induced anti-tumor immunity.45 Taken together, on the basis of our data, we propose that Virus growth and cytotoxicity assays diminished type I IFN response of tumor cells is a prerequisite for Virus replication was assessed by plaque titration of cell culture successful VA7-mediated oncolysis, and that complementing supernatants from infected cells (200 000 cells per well in 6-well plates, MOI = 0.01) collected at indicated time points p.i. and virus was titered by strategies are necessary to potentiate virus-induced anti-tumor plaque assay on VERO(B) cells. For virus cytotoxicity testing, 20 000 cells responses in the setting of intratumoral or intertumoral hetero- were seeded in 150 μl medium volume per well in a 96-well plate. Twenty- geneity. Our work highlights the consequences of clonal virus four hours post seeding, cells were infected with VA7-EGFP virus at resistance and provides a framework for developing improved indicated MOI and 48 h later assessed for viability using Alamar blue

© 2015 Macmillan Publishers Limited Gene Therapy (2015) 65 – 75 Clonal dominance in oncolytic virotherapy JJ Ruotsalainen et al 74 (Invitrogen, Carlsbad, CA, USA) read on Ascent Fluoroscan at 560/590 nm cover slips for ﬂuorescence micrography (Nikon A1R MP, 890 nm, Chiyoda, (Thermo Scientiﬁc, Waltham, MA, USA). Tokyo, Japan).

Supernatant pretreatment experiment SDS-PAGE and western blotting A total of 500 000 cells in 2 ml standard culture medium per well in 6-well Cell lysates from subconfluent VA7-EGFP-infected (MOI = 0.1) and mIFNβ- − plates were infected with VA7-EGFP at MOI = 0.01 and supernatant was treated (100 U ml 1) cells were prepared and run as described previously48 collected 48 h p.i. Virus was removed by 50 kDa size exclusion filtration in two replicas with identical results. Figure 3b presents two blots (Amicon Ultra-4, UFC805008, EMD Millipore, Billerica, MA, USA) and combined from STAT1/STAT1-P and OAS detections (only one of the two supernatant stored at − 70 °C until use. Then, 20 000 cells were seeded in identical actin loading controls is presented in the figure). Primary 96-well plates in 100 μl medium, and after overnight incubation 100 μlof antibodies: Rabbit polyclonal anti-OAS antibody was a kind gift from Ilkka collected supernatant or fresh medium was added into the wells, followed Julkunen (Helsinki, Finland), Rabbit polyclonal anti-STAT1 (BD Transduction 4 h later by 2000 PFU VA7-EGFP in 10 μl per well. Fluorescence Laboratories, San Jose, CA, USA; cat. 610119), Rabbit polyclonal anti-STAT1- micrographs were acquired 24 h p.i. P (Tyr701) (Cell Signaling Technology, Danvers, MA, USA; cat. 91715), monoclonal anti β-actin, C4 (Santa Cruz Biotechnology, Dallas, TX, USA; sc-47778). Secondary antibodies: anti-mouse-cy3 (PA43009) and anti- Interferon pretreatment and VA7 infection assay rabbit-cy5 (PA45011) from GE Healthcare (Little Chalfont, UK). Each treatment and controls were performed in three replicas with identical results and the representative micrographs are shown in Figure 1d. The experiment was carried out on 24-well plates as described Transcriptome sequencing previously17 with or without mouse IFNβ (Sigma Aldrich, St Louis, MO, USA, The RNA-Seq libraries were prepared as earlier described49 out of infected I9032-1VL) pretreatment (1000 U ml − 1, 5 h) coupled with or without VA7- (MOI = 0.01) and uninfected CT26WT and CT26LacZ samples (two replicates EGFP virus infection (MOI = 1 or MOI = 0.01). After final imaging at 48 h p.i., of each) collected 12 h p.i. Method and data analysis details can be found the cells were stained with crystal violet to observe the cytopathic effect. from Supplementary Materials and Methods. Data from this study are available via GEO accession number GSE53001. JAK/STAT pathway inhibition combined with interferon pretreatment and VA7 infection Statistics The experiment was carried out on 48-well plates using three replicas per Kaplan–Meier survival data were analyzed by Log-rank (Mantel-Cox) test, sample type with identical results. VA7 was given at 0 time point, JAK1/ and ELISA and cell cytotoxicity data were analyzed with unpaired two- JAK2 inhibitor Ruxolitinib (Selleckchem, Houston, TX, USA; cat. 1378) in tailed t-test. Tumor volumes were compared by Mann–Whitney U test. fresh medium one day earlier. mIFNβ was given 5 h before VA7. Ruxolitinib − 1 − 1 was used at 1 μM, IFNβ either at 1000 U ml or 100 U ml in standard medium (250 μl per well). CONFLICT OF INTEREST John Bell is co-founder of Jennerex Biotherapeutics and sits on the Board of Directors. fl Antiviral phenotype and gene expression analysis following STAT1 The remaining authors declare no con ict of interest. and LacZ plasmid transfections The experiment was carried out on 12-well plates as three replicas with ACKNOWLEDGEMENTS identical results. CT26LacZ cells were transfected with STAT1 or LacZ We want to thank Diana Schenkwein from A.I. Virtanen Institute for Molecular expression plasmid and infected with 10 PFU VA7-EGFP 4 h later. The Sciences for her expert advice and the staff of the lab animal centers for their infectivity and gene expression were compared with CT26WT cells. professional support. MV-K was funded by the Academy of Finland (grant number Micrograph and western analysis data from one representative sample 125186), Canadian Institutes of Health Research and Terry Fox Foundation, Canada. are presented in Supplementary Figure S4. After the final 48 h imaging JB was supported by an Industrial Fellowship from the Canadian Institutes of Health time point (post transfection), all the samples were lysed for western Research and Terry Fox Foundation (TFF 122868). AH was financially supported by analysis. Detailed description of the experiment can be found in the Academy of Finland (grant number 137958), Emil Aaltonen Foundation, Supplementary Materials and Methods. Medicinska Understödsföreningen Liv och Hälsa, the Finnish Cancer Foundations, the strategic funding for the Cancer Center of the University of Eastern Finland, Animal experiments DPMM graduate school, State Funding for University Hospitals (EVO) and Orion Female Balb/cAnNCrl and Fox Chase SCID mice (5–8-week old) were Foundation. purchased from Charles River Laboratories (Wilmington, MA, USA). Flank tumors were established (2 × 105 cells per injection) in indicated number of mice by intradermal implantation as described previously.47 For mixed REFERENCES tumors, total number of injected cells was kept constant while varying the 1 Russell SJ, Peng KW, Bell JC. Oncolytic virotherapy. Nat Biotechnol 2012; 30: ratios of CT26LacZ to CT26WT cells. Tumor growth rate was monitored 658–670. with a caliper. Tumor take was higher than 90% for CT26LacZ tumors and 2 Miest TS, Cattaneo R. New viruses for cancer therapy: meeting clinical needs. Nat 100% for CT26WT tumors in immunocompetent Balb/c mice and CT26LacZ Rev Microbiol 2014; 12:23–34. tumor take was 100% in SCID mice. Virus or saline (mock) was injected in 3 Noser JA, Mael AA, Sakuma R, Ohmine S, Marcato P, Lee PW et al. The RAS/Raf1/ 10 μl volume with a 30G needle when tumors were clearly palpable MEK/ERK signaling pathway facilitates VSV-mediated oncolysis: implication for the (3–5 mm). Animals were monitored daily and killed upon appearance of defective interferon response in cancer cells. Mol Ther 2007; 15: 1531–1536. severe distress, marked weight loss (420%) or large tumor mass (⩾15 mm 4 Shmulevitz M, Pan LZ, Garant K, Pan D, Lee PW. Oncogenic Ras promotes reovirus on the longest diameter). All experiments were conducted under protocols spread by suppressing IFN-beta production through negative regulation of RIG-I ME-220 and ME-249 and were approved by the local animal welfare signaling. Cancer Res 2010; 70: 4912–4921. committees in Ottawa, Canada and Kuopio, Finland. Virus immunohisto- 5 Saloura V, Wang LC, Fridlender ZG, Sun J, Cheng G, Kapoor V et al. Evaluation of chemistry was carried out as described previously.17 an attenuated vesicular stomatitis virus vector expressing interferon-beta for use in malignant pleural mesothelioma: heterogeneity in interferon responsiveness defines potential efficacy. Hum Gene Ther 2010; 21:51–64. Ex vivo tumor infections 6 Krishnamurthy S, Takimoto T, Scroggs RA, Portner A. Differentially regulated Tumors extracted from CT26WT or CT26LacZ tumor-bearing animals (N =2 interferon response determines the outcome of Newcastle disease virus infection per cell line) were cut to ~ 500 μm thickness (McIlwain tissue chopper, Ted in normal and tumor cell lines. J Virol 2006; 80:5145–5155. Pella Inc., Redding, CA, USA) and placed in cell culture inserts (PICM01250, 7 Berchtold S, Lampe J, Weiland T, Smirnow I, Schleicher S, Handgretinger R et al. EMD Millipore) in 24-well dishes in standard medium. Slices were infected Innate immune defense defines susceptibility of sarcoma cells to measles vaccine immediately with 107 PFU VA7-EGFP and monitored under fluorescence virus-based oncolysis. J Virol 2013; 87: 3484–3501. microscope using × 25 objective. 40 h p.i., slices were washed with 1 × 8 Malilas W, Koh SS, Srisuttee R, Boonying W, Cho IR, Jeong CS et al. Cancer phosphate-buffered saline (10 min), fixed and mounted on slides with upregulated gene 2, a novel oncogene, confers resistance to oncolytic vesicular

Gene Therapy (2015) 65 – 75 © 2015 Macmillan Publishers Limited Clonal dominance in oncolytic virotherapy JJ Ruotsalainen et al 75 stomatitis virus through STAT1-OASL2 signaling. Cancer Gene Ther 2013; 20: 30 Mroz P, Szokalska A, Wu MX, Hamblin MR. Photodynamic therapy of tumors can 125–132. lead to development of systemic antigen-specific immune response. PLoS One 9 Christian SL, Zu D, Licursi M, Komatsu Y, Pongnopparat T, Codner DA et al. 2010; 5: e15194. Suppression of IFN-induced transcription underlies IFN defects generated by 31 Aabo K, Roed H, Vindelov LL, Spang-Thomsen M. A dominated and resistant activated Ras/MEK in human cancer cells. PLoS One 2012; 7: e44267. subpopulation causes regrowth after response to 1,3-bis(2-chloroethyl)-1-nitro- 10 Naik S, Nace R, Federspiel MJ, Barber GN, Peng KW, Russell SJ. Curative one-shot sourea treatment of a heterogeneous small cell lung cancer xenograft in systemic virotherapy in murine myeloma. Leukemia 2012; 26: 1870–1878. nude mice. Cancer Res 1994; 54: 3295–3299. 11 Heikkila JE, Vaha-Koskela MJ, Ruotsalainen JJ, Martikainen MW, Stanford MM, 32 Stojdl DF, Lichty B, Knowles S, Marius R, Atkins H, Sonenberg N et al. Exploiting McCart JA et al. Intravenously administered alphavirus vector VA7 eradicates tumor-specific defects in the interferon pathway with a previously unknown orthotopic human glioma xenografts in nude mice. PLoS One 2010; 5: e8603. oncolytic virus. Nat Med 2000; 6:821–825. 12 Barrett MT, Lenkiewicz E, Evers L, Holley T, Ruiz C, Bubendorf Let al. Clonal 33 Saidi RF, Williams F, Silberberg B, Mittal VK, ReMine SG, Jacobs MJ. Expression of evolution and therapeutic resistance in solid tumors. Front Pharmacol 2013; 4:2. interferon receptors in pancreatic cancer: identification of a novel 13 Vaha-Koskela MJ, Kallio JP, Jansson LC, Heikkila JE, Zakhartchenko VA, Kallajoki MA prognostic factor. Surgery 2006; 139: 743–748. et al. Oncolytic capacity of attenuated replicative semliki forest virus in human 34 Chan SR, Vermi W, Luo J, Lucini L, Rickert C, Fowler AM et al. STAT1-deficient melanoma xenografts in severe combined immunodeficient mice. Cancer Res 2006; mice spontaneously develop estrogen receptor alpha-positive luminal mammary 66:7185–7194. carcinomas. Breast Cancer Res 2012; 14: R16. 14 Maatta AM, Liimatainen T, Wahlfors T, Wirth T, Vaha-Koskela M, Jansson L et al. 35 Simpson JA, Al-Attar A, Watson NF, Scholefield JH, Ilyas M, Durrant LG. Intratu- Evaluation of cancer virotherapy with attenuated replicative Semliki forest virus in moral T cell infiltration, MHC class I and STAT1 as biomarkers of good prognosis in different rodent tumor models. Int J Cancer 2007; 121: 863–870. colorectal cancer. Gut 2010; 59:926–933. 15 Maatta AM, Makinen K, Ketola A, Liimatainen T, Yongabi FN, Vaha-Koskela M et al. 36 Wong LH, Krauer KG, Hatzinisiriou I, Estcourt MJ, Hersey P, Tam ND et al. Replication competent Semliki Forest virus prolongs survival in experimental Interferon-resistant human melanoma cells are deficient in ISGF3 components, lung cancer. Int J Cancer 2008; 123: 1704–1711. STAT1, STAT2, and p48-ISGF3gamma. J Biol Chem 1997; 272: 28779–28785. 16 Ketola A, Hinkkanen A, Yongabi F, Furu P, Maatta AM, Liimatainen T et al. 37 Grudinkin PS, Zenin VV, Kropotov AV, Dorosh VN, Nikolsky NN. EGF-induced Oncolytic Semliki forest virus vector as a novel candidate against unresectable apoptosis in A431 cells is dependent on STAT1, but not on STAT3. Eur J Cell Biol osteosarcoma. Cancer Res 2008; 68:8342–8350. 2007; 86:591–603. 17 Ruotsalainen J, Martikainen M, Niittykoski M, Huhtala T, Aaltonen T, Heikkila J et al. 38 Huang PY, Guo JH, Hwang LH. Oncolytic Sindbis virus targets tumors defective in Interferon-beta sensitivity of tumor cells correlates with poor response to VA7 the interferon response and induces significant bystander antitumor immunity virotherapy in mouse glioma models. Mol Ther 2012; 20: 1529–1539. in vivo. Mol Ther 2012; 20:298–305. 18 Vaha-Koskela MJ, Le Boeuf F, Lemay C, De Silva N, Diallo JS, Cox J et al. Resistance 39 Mittnacht S, Jacobsen H. Selection and characterization of interferon-sensitive to two heterologous neurotropic oncolytic viruses, Semliki Forest virus and cells derived from an interferon-resistant NIH 3T3 line. J Gen Virol 1987; 68(Pt 11): vaccinia virus, in experimental glioma. J Virol 2013; 87: 2363–2366. 2945–2951. 19 Wang M, Bronte V, Chen PW, Gritz L, Panicali D, Rosenberg SA et al. Active 40 Farmer P, Bonnefoi H, Anderle P, Cameron D, Wirapati P, Becette V et al. A stroma- immunotherapy of cancer with a nonreplicating recombinant fowlpox virus related gene signature predicts resistance to neoadjuvant chemotherapy in encoding a model tumor-associated antigen. J Immunol 1995; 154: 4685–4692. breast cancer. Nat Med 2009; 15:68–74. 20 Duca KA, Lam V, Keren I, Endler EE, Letchworth GJ, Novella IS et al. Quantifying 41 Liu YP, Suksanpaisan L, Steele MB, Russell SJ, Peng KW. Induction of antiviral viral propagation in vitro: toward a method for characterization of complex genes by the tumor microenvironment confers resistance to virotherapy. Sci Rep phenotypes. Biotechnol Prog 2001; 17: 1156–1165. 2013; 3: 2375. 21 Ryman KD, Klimstra WB. Host responses to alphavirus infection. Immunol Rev 42 Weichselbaum RR, Ishwaran H, Yoon T, Nuyten DS, Baker SW, Khodarev N et al. 2008; 225:27–45. An interferon-related gene signature for DNA damage resistance is a predictive 22 Barry G, Breakwell L, Fragkoudis R, Attarzadeh-Yazdi G, Rodriguez-Andres J, Kohl A marker for chemotherapy and radiation for breast cancer. Proc Natl Acad Sci USA et al. PKR acts early in infection to suppress Semliki Forest virus production and 2008; 105: 18490–18495. strongly enhances the type I interferon response. J Gen Virol 2009; 90(Pt 6): 43 Cheon H, Stark GR. Unphosphorylated STAT1 prolongs the expression of 1382–1391. interferon-induced immune regulatory genes. Proc Natl Acad Sci USA 2009; 106: 23 Karki S, Li MM, Schoggins JW, Tian S, Rice CM, MacDonald MR. Multiple interferon 9373–9378. stimulated genes synergize with the zinc finger antiviral protein to mediate 44 Li S, Wilkinson M, Xia X, David M, Xu L, Purkel-Sutton A et al. Induction of anti-alphavirus activity. PLoS One 2012; 7: e37398. IFN-regulated factors and antitumoral surveillance by transfected placebo 24 Schoggins JW, Wilson SJ, Panis M, Murphy MY, Jones CT, Bieniasz P et al. A diverse plasmid DNA. Mol Ther 2005; 11:112–119. range of gene products are effectors of the type I interferon antiviral response. 45 Rodriguez-Madoz JR, Zabala M, Alfaro M, Prieto J, Kramer MG, Smerdou C. Nature 2011; 472: 481–485. Short-term intratumoral interleukin-12 expressed from an alphaviral vector is 25 Henrik Gad H, Paulous S, Belarbi E, Diancourt L, Drosten C, Kummerer BM et al. sufficient to induce an efficient antitumoral response against spontaneous The E2-E166K substitution restores Chikungunya virus growth in OAS3 expressing hepatocellular carcinomas. Hum Gene Ther 2014; 25:132–143. cells by acting on viral entry. Virology 2012; 434:27–37. 46 Vaha-Koskela MJ, Tuittila MT, Nygardas PT, Nyman JK, Ehrengruber MU, Renggli M 26 Cho H, Proll SC, Szretter KJ, Katze MG, Gale Jr M, Diamond MS. Differential innate et al. A novel neurotropic expression vector based on the avirulent A7(74) strain immune response programs in neuronal subtypes determine susceptibility to of Semliki Forest virus. J Neurovirol 2003; 9:1–15. infection in the brain by positive-stranded RNA viruses. Nat Med 2013; 19: 47 Bonnotte B, Gough M, Phan V, Ahmed A, Chong H, Martin F et al. Intradermal 458–464. injection, as opposed to subcutaneous injection, enhances immunogenicity and 27 Moerdyk-Schauwecker M, Shah NR, Murphy AM, Hastie E, Mukherjee P, suppresses tumorigenicity of tumor cells. Cancer Res 2003; 63:2145–2149. Grdzelishvili VZ. Resistance of pancreatic cancer cells to oncolytic vesicular 48 Simmons JD, Wollish AC, Heise MT. A determinant of Sindbis virus neurovirulence stomatitis virus: role of type I interferon signaling. Virology 2013; 436:221–234. enables efficient disruption of Jak/STAT signaling. J Virol 2010; 84: 11429–11439. 28 Breitbach CJ, Paterson JM, Lemay CG, Falls TJ, McGuire A, Parato KA et al. Targeted 49 Stender JD, Pascual G, Liu W, Kaikkonen MU, Do K, Spann NJ et al. Control of inflammation during oncolytic virus therapy severely compromises tumor proinflammatory gene programs by regulated trimethylation and demethylation blood flow. Mol Ther 2007; 15: 1686–1693. of histone H4K20. Mol Cell 2012; 48:28–38. 29 Kershaw MH, Hsu C, Mondesire W, Parker LL, Wang G, Overwijk WW et al. 50 Hefti HP, Frese M, Landis H, Di Paolo C, Aguzzi A, Haller O et al. Human MxA Immunization against endogenous retroviral tumor-associated antigens. Cancer protein protects mice lacking a functional alpha/beta interferon system against La Res 2001; 61: 7920–7924. crosse virus and other lethal viral infections. J Virol 1999; 73: 6984–6991.

Supplementary Information accompanies this paper on Gene Therapy website (http://www.nature.com/gt)