Recombinant orthopoxvirus primes colon cancer for checkpoint inhibitor and cross-primes T cells for anti-tumor and anti-viral immunity

Abstract

While it is known that oncolytic viruses can inflame and recruit immune cells to otherwise immunosuppressed tumor microenvironments, the influence of the anti-viral immune response on anti-tumor immunity is less clear across viral platforms and tumor types. CF33 is a recombinant orthopoxvirus backbone effective against colon cancer. We tested derivatives of CF33 with and without immune checkpoint inhibition (anti-PD-L1) in mouse models of colon cancer. Results showed that the efficacy of CF33 backbone with J2R deletion (single-deleted) against colon cancer is not altered by additional deletion of F14.5L in vitro or in vivo. CF33 infection upregulated PD-L1 expression on tumor cells and led to an increased influx of lymphocytes and macrophages in tumors. Also, the levels of active CD8+ (IFNγ+) T cells in the virus-treated tumors were higher than those in control-treated tumors. Furthermore, a combination of CF33-derivatives with anti-PD-L1 resulted in durable tumor regression and long-term survival, resistant to tumor rechallenge. Analysis of immune cells from the treated mice showed that tumor-specific T cell activation occurred more robustly in tumors treated with the virus and that T cells were more strongly activated against the virus than against tumor, in an MHC-I dependent manner. Our findings warrant further studies on the role of cross-priming of T cells against viral and tumor antigens, in the overall success of viroimmunotherapy.

INTRODUCTION

Colon cancer is the third most common cancer in the United States, with approximately 145,000 new cases diagnosed each year, and more than 50,000 deaths in 2019 (1). While surgery with or without chemotherapy can be curative at early stages, almost 50% of patients will ultimately develop liver metastases (2). Surgery is curative for up to 25% of resected patients with colorectal liver metastases (CRLM) at 10 years, (3) however, only a minority of patients are suitable for curative resection and an even smaller proportion of eligible patients get referred for resection (4). Systemic chemotherapies can effectively prolong life for patients with unresectable metastatic disease, but they will ultimately succumb to cancer (5). Thus, there is an urgent need for novel treatment paradigms to eradicate metastatic colorectal cancer and induce durable anti-tumor immunity for improved patient outcomes.

Oncolytic viruses (OV) selectively infect, replicate in, and kill cancer cells. They are either naturally tumor-tropic or are engineered to be tumor-tropic, and can be highly immunogenic across multiple OV families and cancer types (6–8). Poxviruses show potent immunogenicity stemming from both standard anti-viral responses, as would be seen with any viral infection, and from the immunogenic death of infected cancer cells (8). The current understanding in the field suggests that viral infection results in tumor cell lysis, which in turn releases tumor- and virus-associated antigens to spark anti-tumor immunity (8). This ultimately results in a cross-primed memory response against tumor antigens and virally-infected cancer cells (9).

We and others have shown that OVs can activate T cells against tumor cells (10–16). However, many believe that T cell activation after OV infection is predominantly anti-viral and only partly anti-tumoral (17). Some groups are exploiting this by arming OVs with large swaths of tumor antigens (17–19), while others are designing highly replication-competent viruses that can kill cancer cells more efficiently, thereby releasing large quantities of tumor antigens within the tumor microenvironment (TME) (20). We have adopted the latter approach by creating a highly efficient recombinant orthopoxvirus CF33, from the coinfection of nine different orthopoxviruses as previously described (10,21–24). We have also shown that CF33 is robustly immunogenic and efficacious against colon cancer xenografts (22,25), and demonstrated its anti-tumor immune efficacy against triple-negative breast cancer (10). The work herein establishes CF33-derivatives as potent T-cell activators in colon cancer models.

While the field of OV therapy is rapidly gaining momentum, there are two particularly exciting ideas building on viruses as immune activators. First, OVs are robust agents of TME modulation, capable of recruiting and activating immune cells in otherwise immunologically “cold” TMEs devoid of tumor-specific immunity (26). This makes OVs excellent primers for T-cell-based immunotherapies like immune checkpoint inhibitors (ICI) (10,11,27). Second, a very recent article by Rosato et al. elegantly shows that re-stimulating known antiviral immunity can trigger antigen presentation and cytotoxic pathways within infected tumors, which in turn can activate dendritic and NK cells to induce adaptive immunity (28). Thus, it is likely that antiviral immunity can both induce and enhance anti-tumor immunity (9,29).

In this study, we tested if the anti-tumor efficacy of CF33-derivatives against colon cancer could be enhanced by immune checkpoint inhibition (anti-PD-L1). We further characterized the nature of T cell activation following CF33-induced tumor regression. We show that viral infection activates T cells against tumor cells, and in combination with checkpoint inhibition confers durable anti-tumor response. Overall these data suggest that CF33-derivatives in combination with anti-PD-L1 can induce durable anti-tumor immunity paving the way for strategies to enhance immune responses to viral therapy.

MATERIALS AND METHODS:

CF33 Chimerization and hNIS or ΔF14.5 cloning

The chimerization, cloning, competitive selection, and sequence of CF33 backbone virus have been described previously (10,21–23). Insertion of the hNIS expression cassette or firefly luciferase under the control of the vaccinia H5 synthetic early (SE) promoter at the J2R locus has also been described (25,30), as has the deletion of the F14.5L gene (10), and insertion of the anti-PD-L1 transgene at the F14.5L locus (31).

Cell lines and mice

HCT116 (RRID:CVCL_0291), HT-29 (RRID:CVCL_0320), and African green monkey kidney fibroblasts - CV-1 (RRID:CVCL_0229) cell lines were purchased from American Type Culture Collection (ATCC, Manassas, Virginia). All Human colorectal cell lines were maintained in McCoy’s 5A medium (Gibco, Gaithersburg, MD) and CV-1 cells were maintained in Dulbecco’s modified Eagle’s medium (Corning, Corning, NY). Pan02-GFP mouse pancreatic cancer cells were purchased from AntiCancer Inc., San Diego, CA, and cultured in RPMI 1640 medium (Corning, Corning, NY). MC38 and MC38-Luc cells were a kind gift from Dr. Laleh Melstrom’s laboratory (City of Hope, Duarte, CA). MC38 and MC38-Luc cells were maintained in DMEM. All cells were supplemented with 10% fetal bovine serum (FBS) and 1% antibiotic-antimycotic solution, both purchased from Corning (Corning, NY). The cells were maintained in a humidified incubator at 37°C and 5% CO₂. Cell lines were purchased from ATCC and AntiCancer inc, efforts were made not to perform experiments past 15 passages. All cell lines were tested for mycoplasma before each experiment initiation. Mice were maintained in a biosafety containment level 2 facility within our vivarium where the environment was temperature and light controlled with 12-hour light and 12-hour dark cycles, and food and water were ingested ad libitum. All animal experiments were performed with approval of the City of Hope Institutional Animal Care and Use Committee (IACUC). C57Bl/6J mice aged 8–12 weeks old were used for most experiments (Jackson Laboratories, Bar Harbor, ME & Charles River, Wilmington, MA, RRID:IMSR_JAX:000664, RRID:IMSR_CRL:027). Six-week-old Hsd:Athymic Nude-Foxn1nu female mice (Envigo, Indianapolis, IN) were purchased and acclimatized for seven days.

Viral growth assays

HCT-116, HT-29, and MC38-Luc cells were seeded in 6-well plates (Corning, Corning, NY) at 5×10⁵ cells/well and incubated. The following day, cells were counted and infected at an MOI of 0.01 plaque forming unit (PFU) per cell with CF33-hNIS or CF33-hNIS-ΔF14.5 in a total volume of 0.5 mL medium containing 2.5% FBS. After 1 hour, the inoculum was aspirated and 2 mL of fresh medium was added to each well, and plates were returned to the incubator. At the indicated times, cell lysates were collected by scraping and virus titers were determined by standard plaque assay as described previously (10,25).

qRT-PCR

Cells were seeded in 6-well plates at 5×10⁵ cells/well and incubated overnight. The following day, cells were infected at an MOI of 5 with CF33-hNIS-ΔF14.5 and cultured for 24 hours. Total RNA was extracted using the RNeasy Mini kit (Qiagen, Germantown, MD). Next, cDNA was prepared from the total RNA using the QuantiTect Reverse Transcription Kit (Qiagen, Germantown, MD). GAPDH was used as an internal control. Levels of PD-L1 mRNA were assessed using QuantiTect SYBR Green PCR Kit (Qiagen, Germantown, MD). All Primers (Hs_CD274_1_SG QuantiTect Primer, Mm_Pdcd1lg1_1_SG QuantiTect Primer, Hs_GAPDH_1_SG QuantiTect Primer, and Mm_Gapdh_3_SG QuantiTect Primer) were purchased from Qiagen. Real-time PCR was performed on an AB-Quantstudio 3 system (Thermo Fisher Scientific, Waltham, MA). Relative gene expression levels were analyzed by the 2^−ΔΔCt method.

Immunohistochemistry (IHC)

Tumors harvested at the time of sacrifice were fixed with 10% formalin. Subsequently, tumors were paraffin-embedded and 5 μm thick sections were obtained. The slides were deparaffinized followed by heat-mediated antigen-retrieval per manufacturer’s protocol (IHC World, Ellicott City, MD). Tumor sections were then permeabilized with methanol and blocked for 30 minutes with TNB Blocking buffer (PerkinElmer, Waltham, MA). Tumor sections were washed and incubated overnight in a humidified chamber at 4°C with a rabbit anti-vaccinia virus antibody (Abcam, Cambridge, MA, RRID:AB_778768) 1:100 in TNB blocking buffer. The next day, tumor sections were stained with Alexa Fluor-488-conjugated goat anti-rabbit (Abcam, Cambridge, MA, RRID:AB_2630356) for 1 hour at room temperature. Finally, the sections were counterstained with 4′6-diamidino-2-phenylindole (DAPI). IHC for CD8 and Granzyme B was performed by the Pathology Core at City of Hope. Images were obtained using the Nanozoomer 2.0HT digital slide scanner (Hamamatsu Photonics, Hamamatsu City, Shizuoka Pref., Japan) or Ventana iScan HT (Roche, Basel, Switzerland).

Flow cytometry

Single cells were generated from tumors of mice using mouse Tumor Dissociation Kit utilizing GentleMACS dissociator (Miltenyi Biotec, Cologne, Germany). Cells were washed with PBS and stained for different extracellular and intracellular targets. Before staining for targets, cells were stained with LIVE/DEAD Fixable dye (Invitrogen, Carlsbad, CA) in PBS for 30 minutes at 4°C in dark. Next, Fc receptors on the cells were blocked using an anti-CD16/32 antibody (BD Biosciences, Franklin Lakes, NJ, RRID: AB_394657 in FACS buffer(PBS containing 2% FBS) for 10 minutes and then stained for extracellular targets for 30 minutes at 4°C in the dark using the following antibodies: mouse CD3- fluorescein isothiocyanate (FITC) (eBiosciences, San Diego, CA, RRID:AB_2572431), mouse CD11b- allophycocyanin (APC) (Biolegend, San Diego, CA, RRID:AB_389327), mouse PD-L1- phycoerythrin (PE) (Biolegend, San Diego, CA, RRID:AB_2073557), mouse CD45- phycoerythrin (PE) (BD Biosciences, Franklin Lakes, NJ, RRID:AB_394611) or peridinin chlorophyll protein complex (PerCP) (Biolegend, San Diego, RRID:AB_893340), mouse CD8- VioGreen (Miltenyi Biotec, Cologne, Germany RRID:AB_2659495), mouse CD4-APC (Biolegend, San Diego, CA, RRID:AB_389325), and mouse F4/80-PerCP(Biolegend, San Diego, CA, RRID:AB_893496). For intracellular (IFNγ) staining, cells were stained with extracellular markers and then permeabilized using BD Cytofix/Cytoperm Solution Kit according to the manufacturer’s protocol (BD Biosciences). Cells were then incubated with mouse IFNγ-APC (Biolegend, San Diego, CA, RRID:AB_315404) antibody and incubated for 30 minutes at 4°C in dark. Finally, cells were washed twice, resuspended in FACS buffer, and data were acquired using the MACSQuant Analyzer 10 (Miltenyi Biotec, Cologne, Germany). Data were analyzed using the FlowJo software (v10, TreeStar, Ashland, OR).

Tumor models

For the athymic nude mouse models, flank tumors were generated by injecting 2–3 × 10⁶ HCT116 or HT-29 cells in a total of 100 μL PBS containing 50% matrigel for each tumor. When the average tumor size approached 150 mm³, mice were divided into experimental groups and treated with 10⁴ PFU of CF33-hNIS-ΔF14.5 in 50 μL PBS by intratumoral injection.

Flank tumors of MC38-Luc were established using 1–5 × 10⁵ cells in matrigel. Tumor measurements and mouse weight were monitored twice weekly using calipers to calculate tumor volume, V (mm³) = (1/2) x A² x B, where A is the shortest, and B is the longest measurement. Treatment typically occurred when tumors reached 50 – 100 mm³ (approximately 10 days post-cell-injection) following which mice were randomized into treatment groups (n = 4 – 10) such that average tumor volume in each group is similar. In virus with and without anti-PD-L1 experiments, tumors were treated on the same-day with intratumoral (IT) virus and intraperitoneal (IP) anti-PD-L1 injection. Medication dosages were 50 uL PBS IT, 1 × 10⁸ pfu of CF33-hNIS-ΔF14.5 in 50 uL IT, 150 – 200 μg mouse anti-PD-L1 (Bio X Cell, West Lebanon, NH, RRID:AB_10949073) in 100 μL of PBS IT, or 100 μL of PBS IP. Groups included: (i) PBS IT/PBS IP, (ii) PBS IT/mouse anti-PD-L1 IP (m-aPDL1), (iii) CF33-hNIS-ΔF14.5 IT/PBS IP, and (iv) CF33-hNIS-ΔF14.5 IT/m-aPDL1 IP.

Regression and rechallenge experiments for the double-deleted virus were performed using MC38-Luc cells to create unilateral flank tumors on the right flank of each animal. All tumors measuring greater than 130 mm³ at treatment initiation were excluded. Iterations of this experiment were performed with similar results. Representative tumor measurements and survival curves are shown in Figure 3.

Figure 3: A combination of local CF33-hNIS-ΔF14.5 with systemic anti-PD-L1 induces durable tumor regression.

A. A treatment plan is outlined demonstrating that either double-deleted virus alone, anti-PD-L1 alone, or combination treatment (all given on the same day). B. Tumor growth over time is plotted by treatment group with volume in mm³. C. Survival curve demonstrating superior survival in combination group. D. Waterfall plot of largest percent change of individual tumors. Data are representative of experiments repeated in triplicate with similar findings. Stat: Log-rank (Mantel-Cox test) **=p < 0.01

T cell isolation

Mouse splenocytes were obtained from naïve C57BL/6 mice or MC38-Luc tumor-bearing C57BL/6 mice that received indicated treatments. T cells were isolated using the EasySep Mouse T Cell Isolation Kit (StemCell Technologies, Vancouver, BC) according to the manufacturer’s instructions. Freshly isolated mouse T cells were cultured overnight in RPMI medium containing 10% FBS, 10 ng/mL recombinant murine IL-7 (Peprotech, Rocky Hill, NJ), 50 U/mL recombinant human IL-2 (Novartis Oncology, East Hanover, NJ), and 50 μM 2-Mercaptoethanol (Gibco, Gaithersburg, MD). Next day T cells were centrifuged at 300 g for 10 minutes and resuspended in RPMI medium containing 10% FBS. T cells were counted and used for co-culture with MC38-Luc cells.

T cell co-culture and flow cytometry for T-cell activation

2–3 × 10⁴ MC38-Luc cells/well were plated in round-bottom 96-well tissue culture plates (Corning, Corning, NY). After 2 hours, isolated T cells were added to the wells at a ratio of 1:1 (MC38-Luc cells: T cell) with the addition of varying MOIs of the virus. Plates were incubated for 18 hours or 48 hours. After culturing, the expression of T cells activation markers was evaluated by flow cytometry using mouse CD45-PerC (Biolegend, San Diego, CA, RRID:AB_893340), mouse CD8-VioGreen (Miltenyi Biotec, Cologne, Germany RRID:AB_2659495), mouse CD137-PE (Biolegend, San Diego, CA, RRID:AB_2287565) and mouse CD25-PeCy7 (Biolegend, San Diego, CA, RRID:AB_2616762) as described above.

Interferon gamma (IFNγ) ELISA, and major histocompatibility complex (MHC) blockade

Supernatants were collected from T cells and MC38-Luc cells co-culture as described above. For IFNγ ELISA with MHC blockade, 10 μg/mL of anti-mouse MHC Class I (H-2) (Bio X Cell, West Lebanon, NH, RRID:AB_1125537) or anti-mouse MHC Class II (I-A/I-E) (Bio X Cell, West Lebanon, NH, RRID:AB_10949298) were used to block MHC. IFNγ was quantified using mouse IFNγ ELISA Ready-SET-GO! (eBiosciences, San Diego, CA) following manufacturer’s protocol. Plates were read at 450 nm using a spectrophotometer (Tecan Spark 10M, Mannedorf, Switzerland).

Statistical Analysis

Statistical analysis was performed using GraphPad Prism (Version 7.01, La Jolla, CA). Student’s t test or one-way ANOVA were used to evaluate for statistical significance. p < 0.05 was considered significant. Where present in figures, error bars indicate SD or SEM as defined in legends.

RESULTS

Single and double-deleted CF33-derivatives have similar efficacy against colon cancer in vitro and in vivo

We have constructed a recombinant orthopoxvirus (CF33) and modified it as previously described to carry either a single deletion [J2R gene replaced with human sodium iodide symporter (hNIS) transgene] (21,25) or a double-deletion (deletion of J2R and F14.5L genes) to allow for future transgene insertion at F14.5L locus (Fig. 1A) (10,31). Other configurations of double-deletion have included the addition of an anti-PD-L1 transgene as described previously at the F14.5L locus (31). We hypothesized that there would be minimal to no alteration in viral replication between single- and double-deleted CF33-derivatives in human and murine colon cancer cells. Indeed, we observed no attenuation in viral replication in vitro (Fig. 1B) with a single or double deletion. Similarly, we did not observe significant differences in tumor volumes and the survival of nude mouse models with established human xenografts of colon cancer infected with either the single- or double-deleted virus (Fig. 1C–F). These results indicate that single or double-deleted CF33 derivatives have similar efficacy against colon cancer. In previous publications, we have also shown that double-deleted derivatives bearing deleted F14.5L or anti-PD-L1 expression at F14.5L have similar efficacy (31).

Figure 1: Single and double-deleted CF33-derivatives have similar efficacy against colon cancer.

A. CF33-single and CF33-double deleted configurations are shown with single deleted (TK, J2R locus) variant CF33-NIS and double deleted (F14.5L locus) CF33-NIS-ΔF14.5 exemplified. B. Viral growth curve demonstrates replication of CF33-derivatives is unperturbed by additional deletion. Error bars indicate standard deviation C. HCT116 xenograft flank tumors in nude mice were treated with a single dose of intratumoral 1×10⁴ pfu of single deleted (CF33-hNIS, n = 4) or double deleted (CF33-hNIS-aPDL1, n = 4) virus. Data shown are representative of experiments performed at least twice with similar results. Stat = One-way ANOVA D. Survival curves of mice bearing HCT116 xenografts using arrival at 850 mm³ endpoints demonstrate similar curves for single and double deleted treatments. E. HT-29 xenograft flank tumors in nude mice were treated with a single dose of intratumoral 1 × 10⁴ pfu of single deleted (CF33-hNIS, n = 4) or double deleted (CF33-hNIS-aPDL1, n = 3) virus. Data shown are representative of experiments performed at least twice with similar results. Stat = One-way ANOVA F. Survival curves of mice bearing HT-29 xenografts using arrival at 850 mm³ endpoints demonstrate similar curves for single and double deleted treatments. Error bars indicate standard error of the mean. * = p < 0.01; ** = p < 0.001

CF33 infection upregulates PD-L1, and recruits and activates T cells

We have equipped the CF33 virus with transgenes and have previously demonstrated immunogenic cell death in colorectal cancer cells in response to CF33-derivative infection (22). To further understand the ability of our virus to enhance the induction of anti-tumor immunity, we examined PD-L1 expression following viral infection and found that CF33-derivative indeed upregulates PD-L1 mRNA in vitro (Fig. 2A) in human and mouse colon cancer cells.

Figure 2: CF33-derivative infection is immunogenic and activates T cells following infection.

A. Real-time PCR at 24 hours following infection of MC38-Luc (murine) and HCT116 (human) colorectal cancer cells demonstrates an upregulated expression of cellular PD-L1 mRNA. Single cells from tumor at day 2 (B.) following intratumoral infection with 1×10⁸ pfu of CF33-hNIS-ΔF14.5 demonstrate an observed increase in cell-surface PD-L1 also seen along with increased macrophage-dominant infiltration of immune cells, and at day 10 (C.) shows an increase in activated T cells. D. IHC showing tumors with increased CD8+ T cells at 30 days post-infection. Error bars indicate SD. * = p < 0.05; ** = p < 0.01 *** = p < 0.001

Building on previous work, wherein we showed increased immune cell infiltration and activation following CF33-derivative infection in syngeneic models of murine breast and lung cancer (10,21), we hypothesized that similar effects would be observed in the current MC38-Luc model. To investigate this, we injected MC38-Luc tumors in immunocompetent C57Bl/6 mice with a single dose of 1 × 10⁸ PFU of CF33-hNIS-ΔF14.5. Flow cytometric analysis of single cells from the tumors revealed a trend of higher levels of PD-L1 in virus-treated tumors compared to PBS-treated tumors at day 2 post-treatment, however, this difference did not reach statistical significance. The observed increase in PD-L1 levels in virus-treated tumors was abated by day 10 post-infection (Fig. 2B and C). Nevertheless, our data show an increased influx of tumor-infiltrating lymphocytes on day 2 and to a lesser extent on day 10 following virus treatment. We observed a concomitant increase in macrophage and T cell infiltration on day 2, and later an increase in CD8+ T cells (Fig. 2B, C, and D). On day 2, no difference in IFNγ positive T cells was seen between the treatment groups (Fig. 2B), but by day 10, significantly more activated CD8+ T cells were seen in the tumor microenvironment (Fig. 2C).

Combination of CF33-derivatives with anti-PD-L1 shows antigen-specific therapeutic synergy

The above findings suggest that CF33-derivatives can recruit and activate T cells and may thus serve as an effective primer for immune checkpoint inhibitors. Based on the optimal timing of anti-PD-L1 administration suggested by the experience of other groups using vaccinia-based vectors (11,12), we planned a single dose of virus concurrently with systemic anti-PD-L1 (Fig. 3A). We examined mice in four treatment groups, including local CF33-hNIS-ΔF14.5 or systemic anti-PD-L1 alone or in combination, and found that the combination results in optimal and consistent tumor regression (Fig. 3B and D) and long-term survival (Fig. 3C). Mice that experienced complete tumor regression were rechallenged with MC38-Luc cells, which did not show regrowth of tumors except in one mouse (Fig. 4A and B), suggesting that these mice developed anti-tumor immunity. To further decipher the specificity of the anti-tumor immunity, we performed a second round of rechallenge studies in these same mice (Fig. 4C). This time, mice that had been cured by the combination treatment were challenged with MC38-Luc and concurrently with different murine cancer cells (Pan02-GFP) in the opposite flank. All but one mouse developed Pan02-GFP tumors and none re-grew MC38 tumors (Fig. 4D), indicating that the ‘cured’ mice developed tumor-specific immunity.

Figure 4. Rechallenge of mice with tumor regression following CF33-hNIS-ΔF14.5 treatment demonstrates a durable antigen-specific cure.

A. Following the regression of treated MC38-Luc tumors, tumor growth after rechallenge with MC38-Luc cells is plotted by the original treatment group with volume in mm3. B. Survival curve demonstrating long-term survival of all but one mouse in regressed groups versus the death of all mice in naïve group C. Treatment plan showing that mice with regressed tumors after CF33-hNIS-ΔF14.5 treatment were rechallenged with MC38-Luc 2 months post-regression, and then rechallenged again, with MC38 on one flank and Pan02-GFP murine pancreas cells on the other D. Growth of Pan02-GFP but not MC38-Luc allografts suggests tumor antigen-specific anti-tumor immunity. Stat: Log-rank (Mantel-Cox test) *= p < 0.05

T cells isolated from ‘cured’ mice have tumor-specific and virus-specific memory

C57BL/6 mice that experienced complete tumor regression following treatment with anti-PDL1 alone, CF33-derivative alone, and CF33-derivative in combination with anti-PD-L1 were euthanized after 45 days following rechallenge along with age-matched non-tumor bearing mice that never received any treatment. Their spleens were collected and T cells were isolated and co-cultured with MC38-Luc under different treatment conditions as shown in Fig. 5. Treatment conditions of co-cultured MC38-Luc and isolated T cells included: no treatment, anti-PD-L1 alone, CF33-derivative alone, and combination. Flow cytometric analyses were performed at 18 hours (Supplemental fig. 1) and at 48 hours (Fig. 5A) post-infection to examine T cell activation signatures. T cell activation signatures (CD25 and CD137 positivity) from spleens of mice cured by any means were uniformly higher compared with T cells from naïve mice under all co-culture treatment conditions. Furthermore, activation signatures were consistently higher in mice cured by virus treatment either as a single agent or as combination therapy, compared to those cured with anti-PD-L1 treatment alone in all treatment conditions. In the case of CD25+ activation signatures, T cells from animals cured with the virus were significantly more activated in the presence of the virus (Fig. 5A). In the case of CD137+ activation signatures, T cells isolated from virus-treated mice were more highly expressed under all treatment conditions (Fig. 5A). In addition to flow cytometric analyses, we also performed ELISA to quantify IFNγ in the supernatant of T cells co-cultured with the target cells (MC38-Luc) at 18 hours (Supplemental fig. 1) and 48 hours (Fig. 5B) after co-culture. Intriguingly, despite the same levels of T cell surface markers of activation amongst varied co-culture treatment conditions, IFNγ levels were markedly higher in the supernatants of T cells that had been previously exposed to the virus in vivo and were again treated with the virus in vitro. This suggests that a robust systemic anti-viral response may be necessary to maximally prime a tumor microenvironment to induce systemic anti-tumor immunity.

Figure 5 – T cells isolated from cured mice show tumor-specific memory and stronger virus-specific memory.

A. T cells isolated from spleens of naïve mice (never exposed to tumor or treatment), and from mice previously bearing MC38-Luc tumors cured after anti-PD-L1 treatment, virus alone and virus combined with anti-PD-L1 were co-cultured with MC38-Luc cells. They were then treated in vitro with conditions including anti-PD-L1 alone, virus alone, and virus + anti-PD-L1. Flow Cytometric analysis of T cell activation under various co-culture settings was performed demonstrating T cell activation by the original treatment group. T cells from mice previously treated with virus or virus + anti-PD-L1 demonstrated consistently higher activation signatures when co-cultured with MC38-Luc cells under any treatment condition. Stat = one-way ANOVA. No significant difference was seen if not marked by asterisk B. ELISA from supernatants of co-cultures collected at 48 hours and showing IFNγ levels in each treatment group. While some IFNγ secretion is seen when T cells are co-cultured with MC38-Luc, IFNγ production increases significantly following viral infection. C. T cells isolated from spleens of mice cured with virus + anti-PD-L1 were again co-cultured with MC38-Luc cells. This time, co-culture was performed with or without MHC-I blockade. Blocking MHC-1 showed decreased IFNγ production from activated T cells. D. MCH-II blockade leads to no changes in IFNγ production. E. Immunohistochemistry 10 days following intra-tumoral CF33-derivative infection demonstrates the colocalization of T cells with actively infected tumor cells and demonstrates granzyme B staining in areas of infiltrating T cells. Error bars indicate standard deviation. *= p < 0.05; **= p < 0.01 ***= p < 0.001

T cell activation is MHC-I-dependent and induces cytotoxic T cell-based tumor destruction

To further characterize virus-induced T cell activation, we sought to understand the relevance of MHC in viral antigen presentation to T cell activation in co-culture in the treatment groups described above. Again, mice having experienced tumor regression following treatment with virus combined with anti-PD-L1 were rechallenged with MC38-Luc tumors. ‘Cured’ mice demonstrating no tumor re-growth were euthanized after 45 days following rechallenge. T cells isolated from the spleens of these mice were co-cultured with MC38-Luc cells as described above, this time with blockade of either MHC-I or MHC-II. ELISA of co-culture supernatant showed that T cells in the presence of the MHC-I-blocked MC38-Luc cells demonstrated substantially less IFN-γ production than their un-blocked counterparts in response to viral infection (Fig. 5C). MHC-II blockade did not appear to affect IFNγ production. Additional support for the idea that T cell activation from ‘cured’ mice is MHC-I-dependent was found after IHC analysis at 10 days post-infection with a single-deleted CF33-derivative. We observed that CD8+ T cells co-localize with replicating virus (Fig. 5E). Areas of active viral replication and CD8+ T cell predominance were also observed to stain positive for granzyme B (Fig. 5E). Given that granzyme B is produced by cytotoxic T cells responding to foreign antigens presented through MHC-I molecules, these findings further support the idea that T cells found in spleens of ‘cured’ mice are activated in an antigen-specific manner against tumor cells and are possibly more strongly activated against viral antigens. Taken together, these data demonstrate that CF33-hNIS and CF33-hNIS-ΔF14.5 potently activate tumor-specific immunity in the MC38 colon cancer model. Additionally, the strong anti-viral immunity seems to further enhance the tumor-immunity, which was demonstrated by higher levels of CD8+ T cells activation against virus-infected MC38 cells compared to mock-infected cells.

DISCUSSION:

This study confirmed our hypothesis that CF33-derivative infection activates T cells against cancer cells, resulting in synergy with inhibition of the immune checkpoint protein, PD-L1. We found that CF33-hNIS-ΔF14.5 induced durable anti-tumor immunity when combined with anti-PD-L1 treatment against a syngeneic colon cancer model. Durable cure, defined as complete tumor regression and resistance to re-grow tumors upon rechallenge with the same cells, was most often seen when virus treatment was combined with anti-PD-L1 administration. Failure of the ‘cured’ mice to develop MC38 is suggestive of the presence of MC38-specific immunity in those mice. However, the ability of the mice to develop Pan02 tumors but not MC38 tumors when rechallenged concomitantly with MC38 and Pan02 cells provides definitive evidence that the virus alone or in combination with anti-PD-L1 generates tumor-specific immunity in mice that undergo complete tumor regression. While this confirmed a tumor-specific T cell response, our findings showed that memory-based T cell activation is far more robust against virus-infected cancer cells than non-infected cancer cells. We further confirmed that this recognition appears to be MHC-I-dependent. Thus, viral infection with CF33-derivatives appears to cross-prime T cells against both tumor and viral antigens.

Several studies have demonstrated the ability of OVs to sensitize, otherwise resistant, tumors to immunotherapies. While others have demonstrated that anti-tumor immunity induced after combining vaccinia with immune checkpoint inhibition is dependent upon CD8+ T cells, there is a lack of clear evidence implicating viral-specific T cells in the success of oncolytic vaccinia virus. Our study is novel in its quantification of T cell memory-based activation against virus-infected versus uninfected tumor cells. Interestingly, we have shown that when T cells from spleens of cured mice are co-cultured with cancer cells, they have activation signatures that are more robust if the mice were cured by viral treatment (either as a single agent or as part of combination therapy). Still more intriguing, we found that T cells from mice previously cured with the virus (alone or in combination with anti-PD-L1) when co-cultured with MC38-Luc cells and re-treated with the virus in vitro, showed the same activation signatures as T cells that are only in co-culture with MC38-Luc cells, without any virus treatment. However, the activated T cells in the virus treatment groups produce markedly higher levels of IFNγ than their untreated counterparts. This suggests a role for serial virus dosing and further confirms that a robust anti-viral response plays a major role in reprogramming a TME to prime for immunotherapy. This builds on the findings of Rosato et al. who have shown that reactivation of anti-viral T cells can abrogate the growth of poorly immunogenic tumors in mice and render the tumors more susceptible to PD-L1 blockade (28). It also hints at a possible mechanism outside of variable PD-L1 expression as to how viral infection could prime a microenvironment for checkpoint inhibition.

One ongoing debate in the field of oncolytic viroimmunotherapy is whether we should be optimizing tumor cell destruction and then optimizing a host immune response to the resultant exposed tumor antigens, or arming viruses with as many tumor antigens as possible such that a host immune response to virus infection is conveniently also an anti-tumor response (17). Many OV investigators believe that tumor antigen exposure and anti-tumor immunity creates only a fraction of the immune response generated by a viral infection (17,18). In other words, anti-viral immunity is far stronger than anti-tumor immunity. This theory has resulted in the creation of viruses armed with large swaths of tumor antigens (18). Yet, in a heterogeneous tumor, it is unclear if this would be enough to ensure a durable cure. We believe that sufficiently potent viruses that can directly infect tumor cells and thus expose tumor antigens in an immunogenic fashion will be just as effective as antigen-loaded viruses, with the added benefit of creating a tumor-specific cancer vaccine in real-time. This avoids the limitations of more inefficient vectors loaded with some but not all the antigens associated with a given tumor. In this study, we demonstrate that tumor-specific T cell activation induced by CF33 derivatives is, in fact, more strongly activated against viral infection. Further studies are being considered to determine whether a CF33-derivative encoding tumor antigens or other immune stimulants would offer a different pattern of T cell activation.

There are limitations with using murine models, and the durable cure observed in murine models in the current study is not perfectly reflective of human immune responses. Moreover, when examining anti-tumor immunity, it must be noted that MHC-related differences along with species-specific responses to Vaccinia virus make it difficult to predict the extent to which these findings will predict virus function in humans. Furthermore, use of luciferase-tagged cells may stimulate an immune response in and of themselves. Nevertheless, we have clearly and consistently shown durable antigen-specific anti-tumor immunity against a syngeneic mouse model of colon cancer that was most robust when using a combination of CF33-derivative infection and anti-PD-L1 treatment. We have also shown that the anti-viral T cell response is even more robust than the anti-tumor immune response which warrants further study.

While OVs may have a role as first-line single-agent treatments for some already immunogenic malignancies, it seems that the most promising role for OVs in clinical practice will be to recruit immune cells and “heat” immunologically cold tumor microenvironments bereft of activated immune cells (26). Viruses, and particularly orthopoxviruses, can both destroy tumor cells and modulate the surrounding microenvironment to recruit and activate immune cells while exposing tumor antigens to a newly inflamed TME and tumor-infiltrating lymphocytes (18). This study elucidated the strength of CD8+ T cell activation against CF33 infection and tumor cells. Viruses like CF33-derivatives that are both immunogenic agents for tumor destruction and primers for other immunotherapies are likely to find a strong foothold in the climb toward clinical relevance.

Conclusion

CF33-derivatives induce durable anti-tumor immunity when combined with anti-PD-L1 immune checkpoint inhibition. T cell activation occurring in association with viral infection appears to be far more robust against infected tumor cells than against untreated cells. These findings provide support for serial viral dosing strategies and consideration of incorporation of tumor antigens into viral design to enhance anti-tumor T cell activity. Further study of cross-priming between anti-viral and anti-tumor immune responses following viroimmunotherapy will enhance understanding of existing and future clinical treatment paradigms, especially those involving checkpoint inhibition.