Blimp1 Expression Sensitizes Tumor Vasculature to Oncolytic Virus Infection

Article

VEGF-Mediated Induction of PRD1-BF1/Blimp1 Expression Sensitizes Tumor Vasculature to Oncolytic Virus Infection Graphical Abstract

Authors Rozanne Arulanandam, Cory Batenchuk, Fernando A. Angarita, ..., Christina L. Addison, J. Andrea McCart, John C. Bell

Correspondence [email protected]

In Brief Arulanandam et al. show that VEGFR2 signaling in remodelling vessels induces the transcription repressor PRD1-BF1/ Blimp1, which represses the expression of genes involved in type I interferonmediated antiviral signaling, thus allowing oncolytic virus to infect tumor vasculatures to further spread within tumors.

Highlights d

VEGF/VEGFR2 signaling sensitizes endothelial cells to oncolytic virus infection

d

PRD1-BF1 mediates immune suppression in endothelial cells downstream of VEGF

d

PRD1-BF1 is induced in remodelling vessels by chronic or transient VEGF stimulation

d

Infection of tumor vasculature orchestrated by PRD1-BF1 is critical for OV delivery

Arulanandam et al., 2015, Cancer Cell 28, 1–15 August 10, 2015 ª2015 Elsevier Inc. http://dx.doi.org/10.1016/j.ccell.2015.06.009

Please cite this article in press as: Arulanandam et al., VEGF-Mediated Induction of PRD1-BF1/Blimp1 Expression Sensitizes Tumor Vasculature to Oncolytic Virus Infection, Cancer Cell (2015), http://dx.doi.org/10.1016/j.ccell.2015.06.009

Cancer Cell

Article VEGF-Mediated Induction of PRD1-BF1/Blimp1 Expression Sensitizes Tumor Vasculature to Oncolytic Virus Infection Rozanne Arulanandam,1 Cory Batenchuk,1 Fernando A. Angarita,2 Kathryn Ottolino-Perry,2 Sophie Cousineau,1 Amelia Mottashed,1 Emma Burgess,1 Theresa J. Falls,1 Naomi De Silva,1 Jovian Tsang,1 Grant A. Howe,1 Marie-Claude Bourgeois-Daigneault,1 David P. Conrad,1 Manijeh Daneshmand,1 Caroline J. Breitbach,3 David H. Kirn,3 Leda Raptis,6 Subash Sad,7 Harold Atkins,1 Michael S. Huh,1 Jean-Simon Diallo,1 Brian D. Lichty,4 Carolina S. Ilkow,1 Fabrice Le Boeuf,1 Christina L. Addison,1 J. Andrea McCart,2,5 and John C. Bell1,* 1Centre

for Innovative Cancer Therapeutics, Ottawa Hospital Research Institute, Ottawa, ON K1H 8L6, Canada General Research Institute (TGRI), University Health Network, Toronto, ON M5G 2M9, Canada 3SillaJen Biotherapeutics, San Francisco, CA 94111-3380, USA 4McMaster Immunology Research Centre, McMaster University, Hamilton, ON L8S 4K1, Canada 5Department of Surgery, Mount Sinai Hospital, University of Toronto, Toronto, ON M5G 1X5, Canada 6Department of Biomedical and Molecular Sciences, Queen’s University, Kingston, ON K7L 3N6, Canada 7Department of Biochemistry, Microbiology and Immunology, University of Ottawa, Ottawa, ON K1H 8M5, Canada *Correspondence: [email protected] http://dx.doi.org/10.1016/j.ccell.2015.06.009 2Toronto

SUMMARY

Oncolytic viruses designed to attack malignant cells can in addition infect and destroy tumor vascular endothelial cells. We show here that this expanded tropism of oncolytic vaccinia virus to the endothelial compartment is a consequence of VEGF-mediated suppression of the intrinsic antiviral response. VEGF/VEGFR2 signaling through Erk1/2 and Stat3 leads to upregulation, nuclear localization, and activation of the transcription repressor PRD1-BF1/Blimp1. PRD1-BF1 does not contribute to the mitogenic effects of VEGF, but directly represses genes involved in type I interferon (IFN)-mediated antiviral signaling. In vivo suppression of VEGF signaling diminishes PRD1-BF1/Blimp1 expression in tumor vasculature and inhibits intravenously administered oncolytic vaccinia delivery to and consequent spread within the tumor.

INTRODUCTION Engineered oncolytic viruses (OVs) are multi-functional anticancer agents advancing deeply into clinical trials and, in some instances, appear headed for approval (Bell and McFadden, 2014). OVs are designed to exploit unique attributes of the tumor niche, since many of the pathways subverted by the tumor favoring increased survival, immune evasion, deregulated metabolism, and enhanced angiogenesis also compromise the ability of cancer cells to combat virus infection (reviewed in Ilkow et al., 2014). Over the last several years, it has become clear that OVs infect and destroy tumor cells, but in addition prime or re-awaken existing anti-tumor immune responses (Lichty et al., 2014).

Furthermore, some OV platforms have the ability to specifically infect and destroy tumor vasculature (Breitbach et al., 2011a). For instance, systemic delivery of the oncolytic vaccinia virus Pexa-Vec in cancer patients leads to the infection of tumor vasculature and impairment of tumor perfusion in a large percentage of treated patients (Breitbach et al., 2011b, 2013). Tumor neo-vasculature is distinct in several ways from normal blood vessels as a result of chronic exposure to pro-angiogenic cytokines secreted by cancer cells (Chung et al., 2010). The goal of this study was to characterize the signaling events triggered by vascular endothelial growth factor (VEGF) capable of antagonizing cellular antiviral responses, thereby uniquely sensitizing tumor endothelial cells to oncolytic virus infection and destruction.

Significance VEGF secretion within the tumor microenvironment fuels malignant cell growth by promoting angiogenesis. Here we uncover and characterize a signaling cascade initiated by VEGF, leading to direct suppression of the innate antiviral response in activated endothelial cells. Constitutive VEGF signaling, within the context of pathological vascular remodelling, is a critical determinant of oncolytic virus tropism to tumor blood vessels. Thus, pathological VEGF signaling while promoting tumor growth equally provides an opportunity for destruction of tumor vasculature by oncolytic viruses, thereby increasing their therapeutic efficacy. Cancer Cell 28, 1–15, August 10, 2015 ª2015 Elsevier Inc. 1

Please cite this article in press as: Arulanandam et al., VEGF-Mediated Induction of PRD1-BF1/Blimp1 Expression Sensitizes Tumor Vasculature to Oncolytic Virus Infection, Cancer Cell (2015), http://dx.doi.org/10.1016/j.ccell.2015.06.009

(legend on next page)

2 Cancer Cell 28, 1–15, August 10, 2015 ª2015 Elsevier Inc.

Please cite this article in press as: Arulanandam et al., VEGF-Mediated Induction of PRD1-BF1/Blimp1 Expression Sensitizes Tumor Vasculature to Oncolytic Virus Infection, Cancer Cell (2015), http://dx.doi.org/10.1016/j.ccell.2015.06.009

RESULTS VEGF/VEGFR2 Signaling Sensitizes Tumor Vasculature to Infection by Oncolytic Viruses To investigate the sensitivity of tumor vascular endothelial cells to OV infection, we systemically administered an RFPexpressing, mouse-adapted, oncolytic version of vaccinia virus (vvDD-RFP; 1E9 PFU) into C57BL/6 mice bearing MC38 colon carcinoma tumors seeded under a dorsal window chamber. Live confocal tumor microscopy revealed overlap of vvDD-RFP fluorescence and anti-CD31-APC-labeled tumor vasculature (Figure 1A). Quantitative immunohistochemistry (IHC) established that vvDD staining was detected in tumor cells and 25% of CD31+ vessels (Figure 1B). Systemic administration of OV did not result in the infection of normal tissue or vasculature found in the lung, liver, and kidney (Figure 1B). The replication of vvDD-RFP within human umbilical vein endothelial cells (HUVECs) was enhanced in the presence of MC38-conditioned media (MC38 CM). MC38 colon carcinoma cells express and secrete VEGF-A to form highly vascularized tumors in syngeneic C57BL/6 mice (Zhai et al., 1999). A similar increase in vvDD replication was observed in HUVECs treated with recombinant mouse VEGF-A alone (mVEGFA; Figures 1C and 1D). A comparable increase in the infectivity of a GFP-expressing version of the clinical trial candidate Pexa-Vec, WyTK /GFP+ (Breitbach et al., 2011b), was observed in HUVECs grown in conditioned media from human VEGF secreting Caki-1 renal carcinoma cells (Caki-1 CM) or human recombinant VEGF-A (VEGF165; Figures 1E and 1F). The VEGFR inhibitor brivanib (Cai et al., 2008) and the hVEGF-neutralizing monoclonal antibody bevacizumab blocked conditioned media and recombinant VEGF-A enhancement of vaccinia virus replication (Figures 1C–1F) and activation of the VEGF pathway (Figures 1G and S1). Interestingly, the VEGF-mediated enhancement of vaccinia virus replication was observed in other VEGFR2-expressing cell lines including human dermal microvascular endothelial cells (HDMECs) and MDA-MB-231 human breast cancer cells (Figures 1H and 1I). Caki-1 cells, which do not express VEGFR2 (Keyes et al., 2003), did not increase viral output in the presence of VEGF (Figure 1J). These data demon-

strate that direct VEGF/VEGFR2 activation is a prerequisite for sensitizing vascular endothelial cells to oncolytic vaccinia virus replication. VEGF Signaling Promotes a Repression of the Type I Interferon Immune Response in Endothelial Cells To identify pathways downstream of VEGFR2 that render endothelial cells sensitive to oncolytic vaccinia infection, we carried out a global gene expression analysis comparing OVinfected HUVECs treated with VEGF165 against OV-infected, mock-treated controls. Microarray analysis of VEGF165-treated HUVECs infected with WyTK /GFP+ revealed changes in gene expression that were strongly associated with two major gene ontology (GO) classes. Genes upregulated by VEGF165 in the context of infection were strongly associated with GO classifications involved in cell-cycle processes (Figure 2A, left). However, we also found that VEGF treatment of infected HUVECs led to repression of the type I interferon (IFN) response (Figure 2A, right). Consistent with this finding, we found that, in addition to oncolytic vaccinia, other type I IFN-sensitive OV strains, such as herpes simplex virus (HSV), lacking one copy of ɣ34.5, the rhabdoviruses VSVD51, and Maraba virus MG1 (Brun et al., 2010; Stojdl et al., 2003), demonstrated enhanced infectivity in HUVECs when treated with Caki-1 CM or hVEGF165 (Figures S2A–S2C). Moreover, VEGF treatment inhibited the nuclear translocation of signal transducer and activator of transcription-1 (Stat1), which modulates gene expression in response to IFN stimulation (Figure 2B). VEGF pretreatment of HUVECs also was able to counteract the inhibitory effect of IFNa on OV infection (Figure 2C). We performed a parallel analysis on published microarray data (Suehiro et al., 2010) to identify potential mediators of innate antiviral suppression in VEGF-treated, non-infected HUVECs. Our analysis focused on a selective subset of regulators with known activator (n = 569) or repressor (n = 456) functions (Figure S2D). The transcriptional repressor PRD1-BF1, also known as Blimp1 in the mouse (Keller and Maniatis, 1991), emerged as the most interesting candidate due to elevated mRNA expression levels (8-fold increase) upon VEGF treatment, and its immunological function in other cell systems.

Figure 1. VEGF Sensitizes Tumor Vasculature to Infection by Oncolytic Viruses (A) Representative fluorescent confocal micrograph shows subcutaneously allografted MC38-derived tumor infected with oncolytic vaccinia virus (vvDD)-RFP within a window chamber of a C57/BL6 mouse injected with anti-CD31-APC and imaged after 3 days (n = 3). Scale bar represents 100 mm. (B) IHC of oncolytic vvDD-RFP-infected MC38 tumors for CD31 and vvDD from (A). The absolute number of CD31+ blood vessels as well as the presence (+) or absence ( ) of vvDD staining within these blood vessels (see insets) were quantified 3 days post-vvDD. Values represent mean ± SEM per field; lung (n = 3), liver (n = 3), kidney (n = 3), and tumor (n = 3). Scale bar represents 100 mm. (C) Representative fluorescent micrographs of vvDD-RFP-infected (red) HUVECs grown in untreated media (Mock), MC38-conditioned media (MC38 CM), or mouse VEGF-A-supplemented media (mVEGFA). HUVECs were pretreated with VEGFR inhibitor brivanib at 10 nM where indicated. Scale bar represents 50 mm. (D) Viral titers of experiment described in (C). Values represent mean titers ± SEM; *p < 0.05, ***p < 0.001 by Student’s t test. (E) Representative fluorescent micrographs of WyTK /GFP+-infected (green) HUVECs grown in untreated media (Mock), Caki-1-conditioned media (Caki-1 CM), or human VEGF-supplemented media (hVEGF165). HUVECs were treated with or without the VEGF inhibitor bevacizumab at 500 ng/ml where indicated. Scale bar represents 50 mm. (F) Viral titers of experiment described in (E). Values represent mean titers ± SEM; **p < 0.01, ***p < 0.001 by Student’s t test. (G) Immunoblot analysis of VEGFR2 phosphorylation in HUVECs exposed to mock-treated media (untreated), human VEGF165-supplemented media (VEGF165), or Caki-1 CM, with or without bevacizumab (500 ng/ml) as indicated, is shown (n = 3; pooled). (H–J) Viral titers of WyTK /GFP+-infected HDMEC (H), MDA-MB-231 (I), or Caki-1 (J) treated with vehicle (Mock) or VEGF165 along with bevacizumab (500 ng/ml) where indicated. (Inset) VEGFR2 expression in each cell line by immunoblot is shown. Values represent mean titers ± SEM (n = 3); *p < 0.05, **p < 0.01 by Student’s t test. See also Figure S1.

Cancer Cell 28, 1–15, August 10, 2015 ª2015 Elsevier Inc. 3

Please cite this article in press as: Arulanandam et al., VEGF-Mediated Induction of PRD1-BF1/Blimp1 Expression Sensitizes Tumor Vasculature to Oncolytic Virus Infection, Cancer Cell (2015), http://dx.doi.org/10.1016/j.ccell.2015.06.009

Figure 2. VEGF Signaling in Endothelial Cells Promotes a Repression of the Type I IFN Immune Response (A) Gene ontology (GO) analysis of VEGF-responsive genes in WyTK /GFP+-infected endothelial cells. Statistically significant enrichment for GO categories involved in cellular division (left) and type I IFN immune response (right) are shown. (B) Immunoblot analysis of indicated proteins in nuclear and cytoplasmic fractions from HUVECs treated with or without VEGF165 overnight and either mock infected or infected with oncolytic WyTK /GFP+ for 48 hr is shown. (C) Viral titers of WyTK /GFP+-infected HDMEC treated with VEGF165 and IFNa as indicated. Values represent mean titers ± SEM (n = 3); *p < 0.05, **p < 0.01 by Student’s t test. See also Figure S2.

normal human GM38 fibroblasts (Figure S3). Lastly, we confirmed that oncolytic vaccinia virus infection of human primary endothelial cells had no obvious impact on the VEGF-mediated induction of PRD1-BF1 (Figure 3G).

VEGF165 Upregulates the Transcription Repressor PRD1-BF1 in Human Endothelial Cells We hypothesized that VEGF-induced PRD1-BF1 expression could suppress the antiviral response in endothelial cells, as suggested by our GO enrichment analysis. Both qRT-PCR (Figure 3A) and western blot analysis (Figure 3B) revealed that basal expression of PRD1-BF1 is barely detectable in HUVECs, but is induced up to 50-fold in a dose-dependent manner in response to VEGF165. In both HUVECs and HDMECs, the induction of PRD1-BF1 by VEGF165 was abrogated by bevacizumab (Figure 3C) and suppressed by blocking VEGFR signaling using kinase inhibitors, such as brivanib and sunitinib (Figure 3D). Other direct activators of VEGFR2 were capable of inducing PRD1-BF1 expression, including CM from MC38 cells (Figure 3D), the hVEGF121 isoform (Figure 3E), as well as VEGF-E (Figure 3E), a distinct family member that is encoded by the Orf parapoxvirus that does not bind to VEGFR1/Flt-1 (Meyer et al., 1999). Growth factors that induce similar downstream signaling events, such as epidermal growth factor (EGF; Figure 3F), fibroblast growth factor 1 or 2 (FGF1 and FGF2; Figure 3F), or placental growth factor (PlGF; data not shown), which binds to VEGFR1 (Matsumoto and Claesson-Welsh, 2001), were unable to induce PRD1-BF1 expression in either HUVECs or 4 Cancer Cell 28, 1–15, August 10, 2015 ª2015 Elsevier Inc.

PRD1-BF1 Sensitizes HUVECs to Oncolytic Virus Infection by Suppressing the Antiviral Response Transfection of a smart pool of small interfering RNA (siRNA) selectively targeting PRD1-BF1 in HUVECs resulted in >85% knockdown of PRD1-BF1 protein induced by VEGF (Figure 4A). Upon infection with oncolytic vaccinia virus, we observed that PRD1-BF1 knockdown in HUVECs led to a dramatic reduction in GFP transgene expression and viral titers compared to the untreated, and non-targeting, scrambled siControl-transfected cells (Figures 4B and 4C). PRD1-BF1 knockdown in HUVECs had no impact on either cell viability or VEGF-mediated increases in proliferation (Figures 4D and S4A). PRD1-BF1 is a known transcriptional repressor, so we performed microarray analysis on WyTK /GFP+-infected HUVECs pretreated with VEGF165 and transfected with siPRD1-BF1 or Control siRNA. Although PRD1-BF1 knockdown had a marginal impact on the global transcriptome, approximately 60% of genes repressed by VEGF (>2-fold repression, n = 183/22,693) were at least partially rescued by siPRD1-BF1 knockdown (>1.5-fold induction, n = 111/183). There is a significant overlap between the two datasets (p value < 1E 122), indicating that a large proportion of VEGF-mediated transcriptional repressor activities are regulated by PRD1-BF1 under these conditions (Figure 4E). A similar overlap was observed for genes induced >2-fold by VEGF (n = 199) and repressed at least 1.5-fold by PRD1-BF1 knockdown (60/199), but to a much lesser degree (p value < 1E 45). Genes within this dataset include PRD1BF1, which was induced 6-fold by VEGF treatment and repressed 3.6-fold following gene knockdown.

Please cite this article in press as: Arulanandam et al., VEGF-Mediated Induction of PRD1-BF1/Blimp1 Expression Sensitizes Tumor Vasculature to Oncolytic Virus Infection, Cancer Cell (2015), http://dx.doi.org/10.1016/j.ccell.2015.06.009

Figure 3. VEGF Upregulates the Transcription Repressor PRD1-BF1 in Human Endothelial Cells (A) RT-PCR analysis of RNA harvested from HUVECs treated with the indicated concentrations (0–100 ng/ml) of VEGF165 following 24 hr of exposure. PRD1-BF1 expression fold changes were normalized to GAPDH. Values represent fold change ± SEM (n = 4); *p < 0.05, ***p < 0.001 by Student’s t test. (B) Immunoblot analysis of PRD1-BF1 in HUVECs treated with 0–200 ng/ml VEGF165 for 24 hr. Densitometry values for PRD1-BF1 bands were normalized to a-tubulin (n = 3; pooled). (C) RT-PCR analysis of RNA harvested from HUVECs or HDMECs treated with VEGF165 or VEGF165 preincubated 30 min with 500 ng/ml bevacizumab for 24 hr. Expression fold changes were calculated as in (A). Values represent fold change ± SEM (n = 4); *p < 0.05 by Student’s t test. (D) Immunoblot analysis of PRD1-BF1 in HUVECs treated with mVEGFA or MC38 CM. HUVECs were pretreated with DMSO, sunitinib (4 mM), or brivanib (10 nM), where indicated. Densitometry values for PRD1-BF1 bands were normalized to a-tubulin (n = 3; pooled). (E) Immunoblot of HUVECs treated with VEGF165, VEGF-E from Orf virus, or VEGF121 for 48 hr is shown. (F) Immunoblot of HUVECs treated with VEGF165, EGF, FGF1, or FGF2 for 48 hr is shown. (G) HDMECs were treated with VEGF165 with/without 500 ng/ml bevacizumab, where indicated, for 24 hr, and subsequently mock infected or infected with WyTK /GFP+. Cell lysates were harvested 24 hr post-infection (n = 3; pooled) and analyzed by immunoblotting. See also Figure S3.

We examined the gene ontologies associated with the subset of targets that was defined as being repressed >2-fold by VEGF treatment and rescued at least 1.5-fold following PRD1-BF1 knockdown. In Figures 4F and 4G, these genes are significantly associated (2.9–263 enrichment, p value < 1E 8) with antiviral responses, specifically type I IFN signaling. Examples of these

antiviral targets (Figure 4G in red) include RSAD2, OAS1, OAS3, CFH, CXCL10, DDX60, STAT2, and IFIT1. In contrast, of 57 cell-cycle genes induced by VEGF (>2-fold increase), only one gene (MND1) changed upon PRD1-BF1 knockdown (Figure S4A). Furthermore, our dataset reveals a striking overlap between genes repressed by VEGF, or rescued following Cancer Cell 28, 1–15, August 10, 2015 ª2015 Elsevier Inc. 5

Please cite this article in press as: Arulanandam et al., VEGF-Mediated Induction of PRD1-BF1/Blimp1 Expression Sensitizes Tumor Vasculature to Oncolytic Virus Infection, Cancer Cell (2015), http://dx.doi.org/10.1016/j.ccell.2015.06.009

Figure 4. PRD1-BF1 Sensitizes HUVECs to Oncolytic Virus Infection by Suppressing the Antiviral Response (A) Immunoblot analysis of HUVECs transfected with siControl or siPRD1-BF1; then, after 24 hr, treated with VEGF165 where indicated; and harvested 24 hr later is shown (n = 3; pooled sextuplets). (B) HUVECs were transfected with siControl or siPRD1-BF1, 24 hr later treated with or without VEGF165, then infected with oncolytic virus WyTK /GFP+ and titered 48 hr post-infection (n = 3). Values represent mean titers ± SEM; *p < 0.05 by Student’s t test. (C) Representative micrographs of siControl- and siPRD1-BF1-transfected HUVECs infected with WyTK /GFP+ (green) 48 hr post-infection. Scale bars represent 50 mm (top panels) and 20 mm (lower phase contrast panels). (D) Trypan blue cell viability assay of siControl- and siPRD1-BF1-transfected HUVECs treated with VEGF165. Cell viability was measured 24 hr post-VEGF. Values represent mean ± SEM (n = 3); *p < 0.05, ***p < 0.001 by Student’s t test. (E) Histogram illustrates the fold change in gene expression in HUVEC following PRD1-BF1 knockdown for all genes (blue, n = 22,693) or those sub-classified as repressed over 2-fold by VEGF (red, n = 183) after oncolytic vaccinia virus infection. (legend continued on next page)

6 Cancer Cell 28, 1–15, August 10, 2015 ª2015 Elsevier Inc.

Please cite this article in press as: Arulanandam et al., VEGF-Mediated Induction of PRD1-BF1/Blimp1 Expression Sensitizes Tumor Vasculature to Oncolytic Virus Infection, Cancer Cell (2015), http://dx.doi.org/10.1016/j.ccell.2015.06.009

PRD1-BF1 knockdown in the context of vaccinia infection, with those induced by IRF1, a key player in antiviral immunity (Figure S4B; Schoggins et al., 2011). Taken together, these results support our previous observation that PRD1-BF1 knockdown has no effect on VEGF-induced cell division (Figure 4D) and that VEGF-induced mitogenic and immunosuppressive pathways are not coordinately regulated at the transciptional level. These data support the notion that PRD1-BF1 is a master regulator of an immunosuppressive branch of VEGF signaling in endothelial cells. VEGF-Mediated PRD1-BF1 Induction Occurs through a VEGFR2/Erk/Stat3 Signaling Axis We carried out experiments to delineate the molecular effectors that link VEGFR2 to PRD1-BF1 expression in endothelial cells. Stat3 is a known driver of oncogenesis that impacts both innate and adaptive immunity (Yu et al., 2007). Extracellular regulated kinase 1/2 (Erk1/2) is required for maximal transcriptional activity of Stat3 through induction of phosphorylation at serine 727 (Turkson et al., 1999), and was shown to be required for Stat3mediated differentiation of B cells through PRD1-BF1 (Barnes et al., 2012; Yasuda et al., 2011). In primary endothelial cells, treatment with VEGF165 resulted in rapid phosphorylation of both Stat3 and Erk1/2 prior to the induction of PRD1-BF1 expression (Figure 5A). Compounds that block Stat3 phosphorylation and dimerization, such as S3I-201 (Siddiquee et al., 2007), inhibited VEGF-mediated PRD1-BF1 induction, nuclear localization, and sensitization to oncolytic vaccinia virus (Figures 5B–5E and S5A). A comparable effect was observed with the cisplatin analog CPA7 known to inhibit Stat3-DNA binding (data not shown). Similarly, inhibition of Erk1/2 signaling in HUVECs with U0126 blocked PRD1-BF1 expression (Figures 5B–5E, S5B, and S5C). Our data further revealed that PRD1BF1 induction also can be abrogated by these signaling inhibitors in HUVECs chronically treated with VEGF, to mimic the context of pathological angiogenesis within tumor vasculature (Figure S5D). Taken together, these data support a model in which PRD1-BF1-mediated innate immunosuppression in endothelial cells is transduced through the VEGF-A/VEGFR2/Erk/ Stat3 signaling axis. PRD1-BF1/Blimp1 Is Expressed in Tumor Vasculature upon VEGFR2 Signaling We examined PRD1-BF1 levels (Blimp1 in the mouse) within the remodelling vasculature of a tumor using immunohistochemical staining and found that this transcriptional repressor was detected in approximately 80% of the blood vessels in MC38 tumors compared to the normal vasculature of the lungs (34%), liver (19%), and skin (16%) (Figures 6A and 6B). Similar results were obtained in nude mice bearing Caki-1 tumors (not shown). To determine if Blimp1 upregulation observed within MC38 tumor vasculature is triggered by chronic VEGF/VEGFR2 signaling, tumor-bearing mice were treated with B20-4.1.1, a monoclonal

antibody specific to murine VEGF (Liang et al., 2006), or DC101, a neutralizing antibody to VEGFR2 that blocks ligandinduced receptor activation (Witte et al., 1998). Tumors harvested 24 hr post-treatment were stained for Blimp1 (Figure 6C). Quantification of Blimp1-positive vessels revealed a marked reduction upon treatment with either B20-4.1.1 (42% versus 68% for PBS-treated controls) or DC101 (24% versus 76% for Rat IgG-treated controls; Figure 6D). Similar effects were observed when 80 mg/kg sunitinib was administered via oral gavage for 18 hr (32% versus 76% for DMSO-treated controls, Figures S6A and S6B). Blimp1 levels gradually returned to pretreatment levels within 72 hr of administration of either B204.1.1 or DC101, likely due to a rebound in VEGF levels (Figure S6C). In addition, mice were treated i.v. with 5 mg/kg CPA7, a potent Stat3 inhibitor (Assi et al., 2014), or vehicle (diluted DMSO) for 24 hr. Immunohistochemical analysis (Figures 6C and 6D) demonstrated that, similar to VEGF/VEGFR inhibitors, CPA7 led to a significant reduction in Blimp1-positive vessels (25% versus 71% for DMSO-treated controls). Thus, VEGF/ VEGFR/Stat3 signaling significantly contributes to increased Blimp1 levels within MC38 tumor vasculature. Blimp1 Expression in Tumor Vasculature Predicts Sensitivity to Oncolytic Vaccinia Virus Infection of Murine Adenocarcinomas MC38 tumor-bearing C57BL/6 mice were treated with B20-4.1.1 or DC101 intraperitoneally (i.p.) or CPA7 i.v. for 22 hr then administered 1E9 PFU vvDD. Then, 3 days later, mice were sacrificed and tumors were processed for histological analysis. Vaccinia staining revealed a pronounced block in vvDD infection (Figure 7A) when Blimp1 was suppressed (Figures 6C and 6D). Quantification of vaccinia-infected vessels (Figure 7B) revealed a significant reduction following treatment with B20-4.1.1 (15% versus 26% for PBS-treated controls), DC101 (11% versus 36% for Rat IgG-treated controls), or CPA7 (12% versus 34% for diluted DMSO controls), demonstrating the importance of vessel infection to both virus delivery and spread within the tumor. If Blimp1 expression sensitizes tumor vasculature to oncolytic vaccinia virus infection via the suppression of IFN signaling, we predicted that, in animals lacking an IFN response, blocking VEGF signaling would not impede virus replication. We tested this idea using C57/BL6 mice deficient in the receptors for type I IFN (Ifnar1 / mice; Robinson et al., 2012). Following implantation of MC38 tumors into Ifnar / mice, we administered B20-4.1.1 or PBS and then 22 hr later infected with vvDD. IHC and quantification of vaccinia-positive vessels 3 days after virus delivery revealed no significant impact of the anti-VEGF antibody in Ifnar / mice (Figures S7A–S7D, 40% versus 31% for PBS controls or 52% for both treatment conditions after immunofluorescent costaining of sections for CD31 and vaccinia) compared to WTC57BL/6 mice, which have intact IFN signaling (Figures 7A and 7B).

(F) Enriched GO term processes of VEGF-repressed genes that were de-repressed (>1.5-fold) by PRD1-BF1 knockdown upon oncolytic virus infection. The p values in parentheses are the hypergeometric p values following correction for multiple hypothesis testing (Benjamini Hochberg method). (G) Gene expression heatmap of VEGF-repressed subset of genes. Representation of VEGF165-repressed genes (siControl) and their de-repression upon PRD1BF1 knockdown (siPRD1-BF1) is shown. Gene symbols in red are associated with the GO terms depicted in (F). Genes not found in GOrilla are not shown. See also Figure S4.

Cancer Cell 28, 1–15, August 10, 2015 ª2015 Elsevier Inc. 7

Please cite this article in press as: Arulanandam et al., VEGF-Mediated Induction of PRD1-BF1/Blimp1 Expression Sensitizes Tumor Vasculature to Oncolytic Virus Infection, Cancer Cell (2015), http://dx.doi.org/10.1016/j.ccell.2015.06.009

Figure 5. VEGF-Mediated PRD1-BF1 Induction Occurs through a VEGFR2/Erk/Stat3 Signaling Axis (A) Whole-cell lysates were harvested from HUVECs mock treated or treated with VEGF165 at the indicated time points (n = 3; pooled) and then immunoblotted as indicated. (B) HUVECs were treated with DMSO (Mock), bevacizumab (bev, 500 ng/ml), brivanib (10 nM), S3I-201 (75 mM), or U1026 (10 mM) for 30 min prior to stimulation with VEGF165. Whole-cell lysates were harvested 48 hr later (n = 3; pooled) and immunoblotted as indicated. (C) Representative immunofluorescent micrographs of PRD1-BF1 localization in HUVECs grown on glass coverslips and treated as in (B). Scale bar represents 20 mm. (D) HUVECs were treated as in (B) and infected 24 hr later with WyTK /GFP+. Representative fluorescent images of infected HUVECs (green) were captured at 48 hr post-infection. Scale bar represents 50 mm. (E) Viral titers of experiment described in (D). Values represent mean titers ± SEM; *p < 0.05, **p < 0.01 by Student’s t test. See also Figure S5.

We examined the effects of vvDD-mediated infection of blood vessels in the MC38 model. Cleaved caspase-3 and serial H&E staining of MC38 tumor sections revealed an overlap between 8 Cancer Cell 28, 1–15, August 10, 2015 ª2015 Elsevier Inc.

vaccinia infection and areas of cell death and necrosis, including within vessel structures (Figure 7C). Taken together, our data demonstrate the fundamental importance of the VEGF/Blimp1

Please cite this article in press as: Arulanandam et al., VEGF-Mediated Induction of PRD1-BF1/Blimp1 Expression Sensitizes Tumor Vasculature to Oncolytic Virus Infection, Cancer Cell (2015), http://dx.doi.org/10.1016/j.ccell.2015.06.009

Figure 6. PRD1-BF1/Blimp1 Expression in Tumor Vasculature Is Dependent on VEGF-VEGFR2 Signaling (A) IHC for Blimp1 expression in normal tissues and MC38 tumors in C57BL/6 mice. The red plus and minus indicate the presence and absence, respectively, of Blimp1 expression. The black plus indicates the location of tissue vasculature. Scale bars represent 200 mm. (B) Percentage of Blimp1-positive blood vessels in normal and MC38 tumor tissues from (A). Values represent proportional mean ± SEM calculated from biological triplicates; skin (n = 3), liver (n = 3), lung (n = 3), and tumor (n = 6). *p < 0.05, **p < 0.01, ***p < 0.001 by Student’s t test. (C) Blimp1 expression in MC38 tumor vessels from mice treated with B20-4.1.1, DC101, or CPA7. Representative serial sections of MC38 tumors from controltreated (top: PBS [shown], IgG, or DMSO) or drug-treated mice, as indicated, were analyzed by Blimp1 IHC or H&E staining 24 hr after administration. Scale bar represents 200 mm. (D) Percentage of Blimp1-positive blood vessels over total vessels in MC38 tumors from mice treated in (C), as indicated. Values represent proportional mean ± SEM; n = 6 tumors per group; p values were derived by Student’s t test. See also Figure S6.

Cancer Cell 28, 1–15, August 10, 2015 ª2015 Elsevier Inc. 9

Please cite this article in press as: Arulanandam et al., VEGF-Mediated Induction of PRD1-BF1/Blimp1 Expression Sensitizes Tumor Vasculature to Oncolytic Virus Infection, Cancer Cell (2015), http://dx.doi.org/10.1016/j.ccell.2015.06.009

Figure 7. Targeting PRD1-BF1/Blimp1 in Tumor Vasculature Suppresses vvDD Infection of MC38 Tumors (A) Representative H&E and IHC for vaccinia virus images of MC38 tumor sections from mice treated as indicated, for 22 hr prior to i.p. administration of 1E9 PFU vvDD-mCherry. IgG controls (shown) are comparable to PBS or DMSO controls. Tumors were extracted 3 days post-vvDD. Insets representing magnified images of a representative area enable visualization of infected (+) or uninfected ( ) blood vessels. The plus in H&E images indicate tumor vasculature. Scale bar represents 200 mm unless indicated otherwise. (B) Percentage of vvDD-positive blood vessels from MC38 tumor-bearing mice treated in (A). Values represent proportional mean ± SEM, n = 5 tumors per group; p values were derived by Student’s t test. (C) H&E and IHC analysis as indicated of MC38 tumors from mock-treated mice administered 1E9 PFU vvDD-mCherry in (A). Images at the right show higher magnification of the indicated area in the corresponding left images. Arrows in the right indicate representative areas of tumor vessel destruction. Scale bars represent 1 mm (left) and 100 mm (right). See also Figure S7.

10 Cancer Cell 28, 1–15, August 10, 2015 ª2015 Elsevier Inc.

Please cite this article in press as: Arulanandam et al., VEGF-Mediated Induction of PRD1-BF1/Blimp1 Expression Sensitizes Tumor Vasculature to Oncolytic Virus Infection, Cancer Cell (2015), http://dx.doi.org/10.1016/j.ccell.2015.06.009

Figure 8. Blimp1 Mediates Sensitivity to Oncolytic Virus Infection in the WoundHealing Paradigm (A) Representative immunofluorescence images of a monolayer of HUVECs that was mechanically scratched and treated with VEGF165 or VEGF165 with 500 ng/ml bevacizumab, infected with WyTK /GFP+, and imaged 24 hr post-infection. Scale bar represents 200 mm. (B) Representative immunofluorescence images of a monolayer of HUVECs grown on glass coverslips that was mechanically scratched, treated with VEGF165, infected with WyTK /GFP+, and then fixed 24 hr post-infection and stained for PRD1BF1. Scale bar represents 100 mm (left) or 50 mm (right). (C) The bar graph presents the percentage of Blimp1-positive vessels in normal C57BL/6 mouse skin at 1, 3, and 5 days post-wounding. Values represent proportional mean ± SEM, n = 3 tumors per group; *p < 0.05, **p < 0.01 by Student’s t test. The images show representative results of IHC for Blimp1 from animals sacrificed 1 or 5 days postwounding. The red plus and minus indicate the presence and absence, respectively, of Blimp1 expression in wounded vasculature. Scale bar represents 200 mm. (D) IHC of oncolytic vvDD-RFP-infected (day 3) wounded tissue for CD31 and vvDD. The presence (+) of CD31 or vvDD staining in blood vessels was quantified (n = 3). Values represent proportional mean ± SEM. Scale bar represents 100 mm. See also Figure S8.

axis in modulating the permissiveness of tumor blood vessels to vaccinia infection and destruction. The dampening of IFN signaling within tumor vasculature creates a gateway for entry and spread into the tumor of systemically administered oncolytic virus. VEGF Mediates Sensitivity to Oncolytic Vaccinia Virus Infection in Wound-Healing Models There are many parallels in the molecular signals that drive angiogenesis within the tumor microenvironment and those that promote wound healing, with the former being persistent and the latter transient and tightly controlled (Scha¨fer and Werner, 2008). We hypothesized that, similar to a tumor, the remodelling vasculature of a healing wound would be equally susceptible to OV infection. Indeed we found in an in vitro wound-healing assay that VEGF165 promoted vaccinia virus infection selectively along the leading edge of the wound and treatment with bevacizumab was able to abrogate this effect (Figure 8A). In this model, loss of cell-cell contact by mechanical disruption of the quiescent monolayer can overcome the blockade of VEGF receptor signaling moderated by adherens junctions between endothelial cells, leading to migration and wound closure (Dejana, 2004). We compared sparse and confluent HUVEC cultures for their ability to support vaccinia replication following VEGF165 pretreatment. VEGF165 enhanced viral titers by over 10-fold in sub-

confluent cells, but provided little to no enhancement in HUVECs grown to high cell densities (Figures S8A and S8B). Immunofluorescence staining of mechanically wounded cells revealed that PRD1-BF1 expression is localized to the leading edge of a wound upon Vegf165 treatment, in cells that are infected by vaccinia virus (Figure 8B). Scratch assays performed with PRD1-BF1 knockdown cells further indicated that PRD1-BF1 does not impact HUVECs wound closure per se (Figures S8C and S8D). We performed an in vivo assay where dorsal incisions were made on C57BL/6 mice and IHC for Blimp1 was performed on wounded skin collected at 1, 3, and 5 days post-surgery (Figure 8C). Quantification of positively stained vessels revealed a significant and transient increase in Blimp1 at 1 and 3 days post-wounding, returning to baseline levels after 5 days (Figure 8C). We injected vvDD 3 days after wounding and found that 18% of the vessels in the wounded area were infected with vaccinia virus, while no virus was detected in uninjured skin (Figure 8D). Taken together, our data reveal that VEGF-mediated PRD1-BF1 induction and immune suppression occurs within the context of two major types of remodelling vasculature, wounded and malignant. DISCUSSION Oncolytic viruses are engineered to specifically infect malignant cells yet, unexpectedly, also can infect normal vessel-forming Cancer Cell 28, 1–15, August 10, 2015 ª2015 Elsevier Inc. 11

Please cite this article in press as: Arulanandam et al., VEGF-Mediated Induction of PRD1-BF1/Blimp1 Expression Sensitizes Tumor Vasculature to Oncolytic Virus Infection, Cancer Cell (2015), http://dx.doi.org/10.1016/j.ccell.2015.06.009

endothelial cells found within the tumor bed in both animal models and cancer patients. The sensitization of vascular endothelial cells to oncolytic virus infection is dependent upon a signaling network that links the pro-angiogenic growth factor VEGF to anti-viral programs regulated by IFNs. VEGF not only is a key driver of tumor vascularization and metastasis (Leung et al., 1989), but also is thought to play a role in repressing the generation of innate and adaptive immune responses against the tumor (Motz and Coukos, 2011). The transcriptional profiling presented here is consistent with a dual role for VEGF in stimulating vascular endothelial cell proliferation while simultaneously suppressing cellular innate anti-viral immune responses. PRD1BF1 is the key mediator of VEGF suppression of innate immune responses in vascular endothelial cells, but is not necessary for VEGF-driven mitogenic signaling pathways. It was previously demonstrated that PRD1-BF1/Blimp1 competes for the same binding sites recognized by IFN regulatory factors IRF1 and IRF2 in myeloma cells (Doody et al., 2007). Indeed, the PRD1-BF1 consensus binding sequence (A/C)AG(T/C) GAAAG(T/C)(G/T) is almost identical to the IRF-E sequence bound by IRF1, known to transcriptionally activate the type I IFN program in response to most viruses (Schoggins et al., 2011). In the studies described here using human primary endothelial cells, we uncovered a striking overlap between IRF1 and PRD1-BF1 target genes. Our observation that VEGF treatment of endothelial cells antagonizes the anti-viral properties of IFNa supports a model in which VEGF-induced PRD1-BF1 protein antagonizes IRF at the promoters of key antiviral genes. The VEGF-signaling pathway appears to be exploited by particular virus families during natural infections. For example, tissues infected by cytomegalovirus and human herpes virus cause accelerated pathologic angiogenesis due to virusinduced overexpression of cellular VEGF (Masood et al., 2002; Reinhardt et al., 2005). Members of the Parapox family, such as oncolytic Orf (Rintoul et al., 2012; Wise et al., 1999), Pseudocowpox (PCPV) (Ueda et al., 2003), and Bovine papular stomatitis virus (Inder et al., 2007), all encode VEGF homologs. Our observation that anti-VEGF therapies inhibit OV propagation in endothelial cells suggests that anti-angiogenic therapies may reduce virus replication and spread for some types of natural infections. Specific targeting of VEGF, its receptor, or Stat3, lowers PRD1-BF1/Blimp1 expression in tumor vasculature and inhibits the ability of a clinically relevant oncolytic virus to productively infect and destroy blood vessels and subsequently spread throughout the tumor bed. In normal remodelling of vasculature after wounding, it is known that the gene expression signatures are strikingly similar to those found in tumors (Chang et al., 2004; Pedersen et al., 2003; Riss et al., 2006), and neovascularization in healing wounds resembles that of the disorganized, rudimentary-like, permeable vascular networks found in tumors (Wietecha et al., 2013). We found a transient increase in Blimp1 expression following dorsal skin wounding that is co-incident with susceptibility of remodelling normal vessels to OV infection, although this has no impact on the rate of wound healing (J.A. McCart, personal communication). In scratch-based in vitro wound-healing assays, oncolytic vaccinia virus selectively infects the cells expressing PRD1-BF1 at the leading edge. However, in conditions that simulate endothelial interactions in 12 Cancer Cell 28, 1–15, August 10, 2015 ª2015 Elsevier Inc.

normal resting blood vessels, i.e., confluent cells, oncolytic vaccinia virus infectivity of HUVECs is markedly reduced even in the presence of VEGF165. It is likely that tightly associated endothelium of normal blood vessels contributes to the suppression of VEGF signals and resistance to infection. It is perhaps axiomatic that oncolytic virus delivered systemically must exit blood vessels to gain access to the tumor bed or normal tissues and that vascular endothelial cells constitute a significant barrier to virus extravasation. Our data support the idea that normal vasculature does indeed act as a barrier to oncolytic virus infection of normal tissues; however, ill-formed tumor vasculature bathed in VEGF supports oncolytic virus replication and acts as conduit for virus spread from the vascular system into the tumor bed. Oncolytic viruses are a unique platform of therapeutics designed to take advantage of the repressed innate anti-viral state commonly found within cancer cells, but in addition can target vascular endothelial cells bathed in VEGF. Our findings suggest that characterization of the VEGF and PRD1-BF1 status within the tumor may be a valuable marker for the identification of patients who may be more amenable to OV therapy. EXPERIMENTAL PROCEDURES Detailed experimental information is provided in the Supplemental Experimental Procedures. Virus The oncolytic version of Wyeth-based vaccinia virus (WyTK /GFP+) encoding GFP and the mouse-adapted vvDD-RFP or vvDD-mCherry were described previously (Kim et al., 2006) and quantified by plaque assay on U2OS cells. Cell Culture Models HUVECs and HDMECs were purchased from Lonza and grown in endothelial or microvascular endothelial cell growth medium-2 (Lonza). MDA-MB-231, Caki-1, GM38, and MC38 cells were purchased from ATCC and maintained in either DMEM or McCoys 5A medium (for Caki-1) containing 10% fetal bovine serum (FBS). HUVECs, HDMECs, and GM38 cells were not passaged more than three times since resuscitation. Subconfluent cells were synchronized in basal MCDB131 medium (Gibco/ Invitrogen) with 0.5% FBS and pulsed with indicated concentrations of growth factors. Small molecule inhibitors were added at the indicated concentrations, 30 min prior to growth factor treatment. For viable counts, trypan blue exclusion assay was performed on the Vi-CELL counter (Beckman Coulter). For in vitro wound-healing assays, HUVECs were grown on glass coverslips until confluent, treated with VEGF165, scratched with a sterile cell scraper, and infected with WyTK /GFP+ at MOI 0.01 for 24 hr. Fluorescence Microscopy Viral infection of cells was determined by direct detection of GFP or RFP fluorescence 24–72 hr post-infection, as indicated, using an Axiovert S100 fluorescence microscope (Carl Zeiss). For indirect immunofluorescent detection, paraformaldehyde-fixed cells were permeabilized, blocked, incubated with antibodies, and mounted with Prolong anti-fade (Dako) with DAPI. Images were collected using the Zeiss Imager M1 microscope equipped with AxioCam HRm camera (Carl Zeiss). Mouse Models and Treatment Protocols For bio-distribution of vvDD-RFP in normal, tumor, and wounded vasculature, 3E5 MC38 colon carcinoma cells were used in a window chamber model as previously described (Guba et al., 2002; Ottolino-Perry et al., 2014). 1E9 PFU vvDD was administered i.p. and organs harvested after 3 days. For disruption of VEGF/VEGFR signaling, C57BL/6 mice (Charles River Laboratories) were treated i.p. with B20-4.1.1 (Genentech), DC101 (Bio X Cell), Rat IgG control (Sigma), or PBS. CPA7 was a generous gift from Dr. Leda Raptis

Please cite this article in press as: Arulanandam et al., VEGF-Mediated Induction of PRD1-BF1/Blimp1 Expression Sensitizes Tumor Vasculature to Oncolytic Virus Infection, Cancer Cell (2015), http://dx.doi.org/10.1016/j.ccell.2015.06.009

and administered in diluted DMSO, i.v., as described previously (Assi et al., 2014; Littlefield et al., 2008). IHC Organs were fixed in 10% formalin and processed for paraffin embedding. Blocks were serially sectioned and stained with H&E or for specific target antigens: PRDM1/Blimp1 middle region antibody (Aviva Systems Biology), cleaved caspase-3 (New England Biolabs/Cell Signaling Technology), vaccinia polyclonal antibody (quartett; Breitbach et al., 2011a, 2011b, 2013), and CD31 (Santa Cruz Biotechnology), using Vectastain Elite ABC peroxidase kits (rabbit or goat IgG). Slides were imaged using the Aperio ScanScope (Axiovision Technologies) and Aperio ImageScope software. For vessel quantification, the number of Blimp1- or vaccinia-positive vessels were defined as a percentage of total vessels counted (n = 20–100 per tumor section over five fields) from serial H&E- or CD31-stained sections. Western Blotting Cell lysis and fractionation experiments were performed as previously described (Vultur et al., 2004). Following protein determination by Bradford assay (Bio-Rad Protein Assay Solution), 10–20 mg of clarified cell extract was analyzed using the NuPAGE SDS-PAGE Gel system (Invitrogen). Blots were cut into strips and probed with specific antibodies. Bands were visualized using Supersignal West Pico Chemiluminescent substrate (Thermo Scientific Pierce). qRT-PCR RNA extraction was performed from cells grown in six-well dishes using QIAGEN QiaShredder columns and the QIAGEN RNeasy kit; 2 mg RNA was converted to cDNA with Superscript II Reverse Transcriptase (Invitrogen). Real-time PCR reactions were performed with the QuantiTect SYBR Green PCR kit (QIAGEN) on a Rotor-gene RG-3000 (Corbett Research). Optimal threshold and reaction efficiency were determined using the Rotor-gene software. Melt curves for PRDM1 exhibited a single peak, indicating specific amplification, which also was confirmed by agarose gel. Ct values were determined using the Rotor-gene software at the optimal threshold. Gene expression relative to GAPDH was calculated using the method described previously (Pfaffl et al., 2004). Fold induction or PRDM1 was calculated relative to the untreated control. Primers were designed using Primer 3 version 4.0. PRD1-BF1 Knockdown for Affymetrix Analysis HUVECs in six-well dishes were transfected with SMARTpool ONTARGETplus PRDM1 siRNA, an siRNA scrambled control, or oligofectamine (Invitrogen) alone. Then, 24 hr later, cells were treated with VEGF overnight and infected, as described; 24 hr later, cells from quadruplicate wells were collected and RNA extracted, diluted to a concentration of 100 ng/ml, and pooled at equal ratios. The quality of the final samples was confirmed using an Agilent Bioanalyzer. Samples were hybridized to an Affymetrix human gene 2.0 ST array according to the manufacturer’s instructions. Probeset filtering and data normalization were performed using Affymetrix Expression console. Normalized data were filtered by selecting the probeset with the highest average signal for a single ensemble gene identifier and removing all genes with a signal in the bottom quartile of the dataset. Gene ontology enrichments were performed using GOrilla (Eden et al., 2009). Statistical Analysis Statistical analyses were performed on Microsoft Excel statistical analysis software package. All p values considered significant were p < 0.05. All p values reported for the GO term analysis were calculated using the hypergeometric model following correction for multiple hypothesis testing using the Benjamini and Hochberg method. Animal Approval Animal experiments were approved and performed in accordance with institutional guidelines review board for animal care (University of Ottawa ethical board and Animal Care Centre, University Health Network). Survival endpoint criteria were defined as severe abdominal distension, cachexia, or development of subcutaneous injection site tumor >1.5 cm in diameter.

ACCESSION NUMBERS The accession number for the microarray data reported in this paper is ArrayExpress: E-MTAB-2963. SUPPLEMENTAL INFORMATION Supplemental Information includes Supplemental Experimental Procedures and eight figures and can be found with this article online at http://dx.doi. org/10.1016/j.ccell.2015.06.009. AUTHOR CONTRIBUTIONS Conception and design, R.A., C.B., J.C.B., B.D.L., J.A.M., C.L.A., H.A., C.J.B., and D.H.K.; in vitro experiments, R.A., C.B., S.C., C.S.I., A.M., E.B., and G.A.H.; in vivo experiments, R.A., T.J.F., F.A.A., F.L.B., M.-C.B.-D., K.O.-P., J.T., N.D.S., L.R., and S.S.; writing, review, and revision of the manuscript, R.A., M.S.H., C.B., M.D., J.C.B., D.P.C., F.L.B., J.-S.D., and C.L.A. ACKNOWLEDGMENTS J.C.B. is supported by a Terry Fox Foundation (TFF) Grant, the Canadian Cancer Society Research Institute (CCSRI), the Ontario Institute for Cancer Research (OICR), the Canadian Institute for Health Research (CIHR), the Ottawa Regional Cancer Foundation, and the Ottawa Hospital Foundation. R.A. was supported by a Mitacs Elevate Industrial Fellowship. C.B. was supported by a Natural Sciences and Engineering Research Council of Canada (NSERC) scholarship. S.C. was supported by an NSERC award. L.R. is funded by NSERC, CIHR, and the Canadian Breast Cancer Foundation (CBCF). S.S. is supported by CIHR. C.S.I. was supported by a fellowship from the Alberta Innovates Health Solutions. J.-S.D. is supported by TFF, CIHR, and CCSRI. B.D.L. is supported by CIHR and OICR. J.A.M. is funded by OICR, CCSRI, and CIHR. We thank the J.C.B., J.-S.D., H.A., and Rebecca Auer labs for constructive critique and support. Thanks to M.A. Christine Pratt and Miguel A. Cabrita for guidance, Ellias Horner and Alexandre Blais for imaging, Christiano Souza for assistance with animal injections, Ana Giassi and Zaida Ticas for prompt histology services, and Pamela S. Lagali for confocal imaging. C.J.B. is an employee of Sillajen Biotherapeutics. J.C.B., D.H.K., and H.A. were co-founders of Jennerex Biotherapeutics. J.C.B. and B.D.L. are co-founders of Turnstone Biologics. Received: September 10, 2014 Revised: March 30, 2015 Accepted: June 23, 2015 Published: July 23, 2015 REFERENCES Assi, H.H., Paran, C., VanderVeen, N., Savakus, J., Doherty, R., Petruzzella, E., Hoeschele, J.D., Appelman, H., Raptis, L., Mikkelsen, T., et al. (2014). Preclinical characterization of signal transducer and activator of transcription 3 small molecule inhibitors for primary and metastatic brain cancer therapy. J. Pharmacol. Exp. Ther. 349, 458–469. Barnes, N.A., Stephenson, S., Cocco, M., Tooze, R.M., and Doody, G.M. (2012). BLIMP-1 and STAT3 counterregulate microRNA-21 during plasma cell differentiation. J. Immunol. 189, 253–260. Bell, J., and McFadden, G. (2014). Viruses for tumor therapy. Cell Host Microbe 15, 260–265. Breitbach, C.J., De Silva, N.S., Falls, T.J., Aladl, U., Evgin, L., Paterson, J., Sun, Y.Y., Roy, D.G., Rintoul, J.L., Daneshmand, M., et al. (2011a). Targeting tumor vasculature with an oncolytic virus. Mol. Ther. 19, 886–894. Breitbach, C.J., Burke, J., Jonker, D., Stephenson, J., Haas, A.R., Chow, L.Q.M., Nieva, J., Hwang, T.-H., Moon, A., Patt, R., et al. (2011b). Intravenous delivery of a multi-mechanistic cancer-targeted oncolytic poxvirus in humans. Nature 477, 99–102. Breitbach, C.J., Arulanandam, R., De Silva, N., Thorne, S.H., Patt, R., Daneshmand, M., Moon, A., Ilkow, C., Burke, J., Hwang, T.-H., et al. (2013).

Cancer Cell 28, 1–15, August 10, 2015 ª2015 Elsevier Inc. 13

Please cite this article in press as: Arulanandam et al., VEGF-Mediated Induction of PRD1-BF1/Blimp1 Expression Sensitizes Tumor Vasculature to Oncolytic Virus Infection, Cancer Cell (2015), http://dx.doi.org/10.1016/j.ccell.2015.06.009

Oncolytic vaccinia virus disrupts tumor-associated vasculature in humans. Cancer Res. 73, 1265–1275.

vascular endothelial growth factor/vascular endothelial growth factor receptor. Am. J. Pathol. 160, 23–29.

Brun, J., McManus, D., Lefebvre, C., Hu, K., Falls, T., Atkins, H., Bell, J.C., McCart, J.A., Mahoney, D., and Stojdl, D.F. (2010). Identification of genetically modified Maraba virus as an oncolytic rhabdovirus. Mol. Ther. 18, 1440–1449.

Matsumoto, T., and Claesson-Welsh, L. (2001). VEGF receptor signal transduction. Sci. STKE 2001, re21.

Cai, Z.W., Zhang, Y., Borzilleri, R.M., Qian, L., Barbosa, S., Wei, D., Zheng, X., Wu, L., Fan, J., Shi, Z., et al. (2008). Discovery of brivanib alaninate ((S)-((R)-1(4-(4-fluoro-2-methyl-1H-indol-5-yloxy)-5-methylpyrrolo[2,1-f][1,2,4]triazin-6yloxy)propan-2-yl)2-aminopropanoate), a novel prodrug of dual vascular endothelial growth factor receptor-2 and fibroblast growth factor receptor-1 kinase inhibitor (BMS-540215). J. Med. Chem. 51, 1976–1980. Chang, H.Y., Sneddon, J.B., Alizadeh, A.A., Sood, R., West, R.B., Montgomery, K., Chi, J.-T., van de Rijn, M., Botstein, D., and Brown, P.O. (2004). Gene expression signature of fibroblast serum response predicts human cancer progression: similarities between tumors and wounds. PLoS Biol. 2, E7.

Meyer, M., Clauss, M., Lepple-Wienhues, A., Waltenberger, J., Augustin, H.G., Ziche, M., Lanz, C., Bu¨ttner, M., Rziha, H.J., and Dehio, C. (1999). A novel vascular endothelial growth factor encoded by Orf virus, VEGF-E, mediates angiogenesis via signalling through VEGFR-2 (KDR) but not VEGFR-1 (Flt-1) receptor tyrosine kinases. EMBO J. 18, 363–374. Motz, G.T., and Coukos, G. (2011). The parallel lives of angiogenesis and immunosuppression: cancer and other tales. Nat. Rev. Immunol. 11, 702–711. Ottolino-Perry, K., Tang, N., Head, R., Ng, C., Arulanandam, R., Angarita, F.A., Acuna, S.A., Chen, Y., Bell, J., Dacosta, R.S., and McCart, J.A. (2014). Tumor vascularization is critical for oncolytic vaccinia virus treatment of peritoneal carcinomatosis. Int. J. Cancer 134, 717–730.

Dejana, E. (2004). Endothelial cell-cell junctions: happy together. Nat. Rev. Mol. Cell Biol. 5, 261–270.

Pedersen, T.X., Leethanakul, C., Patel, V., Mitola, D., Lund, L.R., Danø, K., Johnsen, M., Gutkind, J.S., and Bugge, T.H. (2003). Laser capture microdissection-based in vivo genomic profiling of wound keratinocytes identifies similarities and differences to squamous cell carcinoma. Oncogene 22, 3964–3976.

Doody, G.M., Stephenson, S., McManamy, C., and Tooze, R.M. (2007). PRDM1/BLIMP-1 modulates IFN-gamma-dependent control of the MHC class I antigen-processing and peptide-loading pathway. J. Immunol. 179, 7614– 7623.

Pfaffl, M.W., Tichopad, A., Prgomet, C., and Neuvians, T.P. (2004). Determination of stable housekeeping genes, differentially regulated target genes and sample integrity: BestKeeper–Excel-based tool using pair-wise correlations. Biotechnol. Lett. 26, 509–515.

Eden, E., Navon, R., Steinfeld, I., Lipson, D., and Yakhini, Z. (2009). GOrilla: a tool for discovery and visualization of enriched GO terms in ranked gene lists. BMC Bioinformatics 10, 48.

Reinhardt, B., Schaarschmidt, P., Bossert, A., Lu¨ske, A., Finkenzeller, G., Mertens, T., and Michel, D. (2005). Upregulation of functionally active vascular endothelial growth factor by human cytomegalovirus. J. Gen. Virol. 86, 23–30.

Guba, M., von Breitenbuch, P., Steinbauer, M., Koehl, G., Flegel, S., Hornung, M., Bruns, C.J., Zuelke, C., Farkas, S., Anthuber, M., et al. (2002). Rapamycin inhibits primary and metastatic tumor growth by antiangiogenesis: involvement of vascular endothelial growth factor. Nat. Med. 8, 128–135.

Rintoul, J.L., Lemay, C.G., Tai, L.-H., Stanford, M.M., Falls, T.J., de Souza, C.T., Bridle, B.W., Daneshmand, M., Ohashi, P.S., Wan, Y., et al. (2012). ORFV: a novel oncolytic and immune stimulating parapoxvirus therapeutic. Mol. Ther. 20, 1148–1157.

Ilkow, C.S., Swift, S.L., Bell, J.C., and Diallo, J.-S. (2014). From scourge to cure: tumour-selective viral pathogenesis as a new strategy against cancer. PLoS Pathog. 10, e1003836.

Riss, J., Khanna, C., Koo, S., Chandramouli, G.V.R., Yang, H.H., Hu, Y., Kleiner, D.E., Rosenwald, A., Schaefer, C.F., Ben-Sasson, S.A., et al. (2006). Cancers as wounds that do not heal: differences and similarities between renal regeneration/repair and renal cell carcinoma. Cancer Res. 66, 7216–7224.

Chung, A.S., Lee, J., and Ferrara, N. (2010). Targeting the tumour vasculature: insights from physiological angiogenesis. Nat. Rev. Cancer 10, 505–514.

Inder, M.K., Ueda, N., Mercer, A.A., Fleming, S.B., and Wise, L.M. (2007). Bovine papular stomatitis virus encodes a functionally distinct VEGF that binds both VEGFR-1 and VEGFR-2. J. Gen. Virol. 88, 781–791. Keller, A.D., and Maniatis, T. (1991). Identification and characterization of a novel repressor of beta-interferon gene expression. Genes Dev. 5, 868–879. Keyes, K.A., Mann, L., Cox, K., Treadway, P., Iversen, P., Chen, Y.-F., and Teicher, B.A. (2003). Circulating angiogenic growth factor levels in mice bearing human tumors using Luminex Multiplex technology. Cancer Chemother. Pharmacol. 51, 321–327. Kim, J.H., Oh, J.Y., Park, B.H., Lee, D.E., Kim, J.S., Park, H.E., Roh, M.S., Je, J.E., Yoon, J.H., Thorne, S.H., et al. (2006). Systemic armed oncolytic and immunologic therapy for cancer with JX-594, a targeted poxvirus expressing GM-CSF. Mol. Ther. 14, 361–370. Leung, D.W., Cachianes, G., Kuang, W.J., Goeddel, D.V., and Ferrara, N. (1989). Vascular endothelial growth factor is a secreted angiogenic mitogen. Science 246, 1306–1309. Liang, W.-C., Wu, X., Peale, F.V., Lee, C.V., Meng, Y.G., Gutierrez, J., Fu, L., Malik, A.K., Gerber, H.-P., Ferrara, N., and Fuh, G. (2006). Cross-species vascular endothelial growth factor (VEGF)-blocking antibodies completely inhibit the growth of human tumor xenografts and measure the contribution of stromal VEGF. J. Biol. Chem. 281, 951–961. Lichty, B.D., Breitbach, C.J., Stojdl, D.F., and Bell, J.C. (2014). Going viral with cancer immunotherapy. Nat. Rev. Cancer 14, 559–567. Littlefield, S.L., Baird, M.C., Anagnostopoulou, A., and Raptis, L. (2008). Synthesis, characterization and Stat3 inhibitory properties of the prototypical platinum(IV) anticancer drug, [PtCl3(NO2)(NH3)2] (CPA-7). Inorg. Chem. 47, 2798–2804. Masood, R., Cesarman, E., Smith, D.L., Gill, P.S., and Flore, O. (2002). Human herpesvirus-8-transformed endothelial cells have functionally activated

14 Cancer Cell 28, 1–15, August 10, 2015 ª2015 Elsevier Inc.

Robinson, N., McComb, S., Mulligan, R., Dudani, R., Krishnan, L., and Sad, S. (2012). Type I interferon induces necroptosis in macrophages during infection with Salmonella enterica serovar Typhimurium. Nat. Immunol. 13, 954–962. Scha¨fer, M., and Werner, S. (2008). Cancer as an overhealing wound: an old hypothesis revisited. Nat. Rev. Mol. Cell Biol. 9, 628–638. Schoggins, J.W., Wilson, S.J., Panis, M., Murphy, M.Y., Jones, C.T., Bieniasz, P., and Rice, C.M. (2011). A diverse range of gene products are effectors of the type I interferon antiviral response. Nature 472, 481–485. Siddiquee, K., Zhang, S., Guida, W.C., Blaskovich, M.A., Greedy, B., Lawrence, H.R., Yip, M.L.R., Jove, R., McLaughlin, M.M., Lawrence, N.J., et al. (2007). Selective chemical probe inhibitor of Stat3, identified through structure-based virtual screening, induces antitumor activity. Proc. Natl. Acad. Sci. USA 104, 7391–7396. Stojdl, D.F., Lichty, B.D., tenOever, B.R., Paterson, J.M., Power, A.T., Knowles, S., Marius, R., Reynard, J., Poliquin, L., Atkins, H., et al. (2003). VSV strains with defects in their ability to shutdown innate immunity are potent systemic anti-cancer agents. Cancer Cell 4, 263–275. Suehiro, J., Hamakubo, T., Kodama, T., Aird, W.C., and Minami, T. (2010). Vascular endothelial growth factor activation of endothelial cells is mediated by early growth response-3. Blood 115, 2520–2532. Turkson, J., Bowman, T., Adnane, J., Zhang, Y., Djeu, J.Y., Sekharam, M., Frank, D.A., Holzman, L.B., Wu, J., Sebti, S., and Jove, R. (1999). Requirement for Ras/Rac1-mediated p38 and c-Jun N-terminal kinase signaling in Stat3 transcriptional activity induced by the Src oncoprotein. Mol. Cell. Biol. 19, 7519–7528. Ueda, N., Wise, L.M., Stacker, S.A., Fleming, S.B., and Mercer, A.A. (2003). Pseudocowpox virus encodes a homolog of vascular endothelial growth factor. Virology 305, 298–309.

Please cite this article in press as: Arulanandam et al., VEGF-Mediated Induction of PRD1-BF1/Blimp1 Expression Sensitizes Tumor Vasculature to Oncolytic Virus Infection, Cancer Cell (2015), http://dx.doi.org/10.1016/j.ccell.2015.06.009

Vultur, A., Cao, J., Arulanandam, R., Turkson, J., Jove, R., Greer, P., Craig, A., Elliott, B., and Raptis, L. (2004). Cell-to-cell adhesion modulates Stat3 activity in normal and breast carcinoma cells. Oncogene 23, 2600–2616. Wietecha, M.S., Cerny, W.L., and DiPietro, L.A. (2013). Mechanisms of vessel regression: toward an understanding of the resolution of angiogenesis. Curr. Top. Microbiol. Immunol. 367, 3–32. Wise, L.M., Veikkola, T., Mercer, A.A., Savory, L.J., Fleming, S.B., Caesar, C., Vitali, A., Makinen, T., Alitalo, K., and Stacker, S.A. (1999). Vascular endothelial growth factor (VEGF)-like protein from orf virus NZ2 binds to VEGFR2 and neuropilin-1. Proc. Natl. Acad. Sci. USA 96, 3071–3076. Witte, L., Hicklin, D.J., Zhu, Z., Pytowski, B., Kotanides, H., Rockwell, P., and Bo¨hlen, P. (1998). Monoclonal antibodies targeting the VEGF receptor-2 (Flk1/

KDR) as an anti-angiogenic therapeutic strategy. Cancer Metastasis Rev. 17, 155–161. Yasuda, T., Kometani, K., Takahashi, N., Imai, Y., Aiba, Y., and Kurosaki, T. (2011). ERKs induce expression of the transcriptional repressor Blimp-1 and subsequent plasma cell differentiation. Sci. Signal. 4, ra25. Yu, H., Kortylewski, M., and Pardoll, D. (2007). Crosstalk between cancer and immune cells: role of STAT3 in the tumour microenvironment. Nat. Rev. Immunol. 7, 41–51. Zhai, Y., Ni, J., Jiang, G.W., Lu, J., Xing, L., Lincoln, C., Carter, K.C., Janat, F., Kozak, D., Xu, S., et al. (1999). VEGI, a novel cytokine of the tumor necrosis factor family, is an angiogenesis inhibitor that suppresses the growth of colon carcinomas in vivo. FASEB J. 13, 181–189.

Cancer Cell 28, 1–15, August 10, 2015 ª2015 Elsevier Inc. 15