Intratumoral PD-1-directed IL-2 expression via oncolytic vaccinia virus elicits superior antitumor effects with enhanced safety

Abstract

Background

Cancer immunotherapy has revolutionized treatment, offering new hope for patients with previously hard-to-treat cancers. However, many solid tumors are classified as non-inflamed, meaning they are less responsive to immunotherapies. This underscores the need for novel strategies to transform these non-inflamed tumors into immune-inflamed ones. Oncolytic viruses, particularly when engineered, hold promise for stimulating immune responses and transforming these tumors. Cytokines, in particular, are promising for arming these viruses. Immune checkpoint inhibition, especially targeting the programmed cell death protein-1 (PD-1)/programmed death-ligand 1 (PD-L1) pathway, has been groundbreaking. In this study, we hypothesize that an oncolytic vaccinia virus (oVV) expressing a fusion protein of the PD-1 ectodomain and interleukin (IL)-2 (vvDD-PD-1ecto-R-IL-2) could mimic the combined effects of anti-PD-L1 antibodies and IL-2, transforming non-inflamed tumors and enhancing antitumor effects while minimizing the toxicity associated with systemic IL-2.

Methods

We generated the vvDD-PD-1ecto-R-IL-2 oncolytic virus and confirmed transgene expression through reverse transcription quantitative PCR (RT-qPCR), ELISA and flow cytometry. The antitumor efficacy and safety of vvDD-PD-1ecto-R-IL-2 were evaluated in murine tumor models. The underlying mechanisms of action were investigated using RT-qPCR, ELISA and flow cytometry.

Results

Our data demonstrate that the vvDD-PD-1ecto-R-IL-2 elicited significant antitumor effects, though the PD-1 ectodomain did not fully mimic the function of anti-PD-L1 antibodies, likely due to the elevated PD-L1 levels following oVV infection. However, PD-1 successfully directed IL-2 to associate with the cell membrane via PD-1ecto-R-IL-2/PD-L1 interaction, minimizing systemic toxicity. The accumulated IL-2 in the tumor microenvironment led to an increase in antitumor factors, including interferon (IFN)-γ, granzyme B, and perforin, while reducing protumor factors such as vascular endothelial growth factor and prostaglandin-endoperoxide synthase 2. Furthermore, treatment with the vvDD-PD-1ecto-R-IL-2 promoted the expansion of tumor-infiltrating stem-like CD8⁺ T cells (CD8⁺TCF-1⁺), as well as functional CD8⁺ T-cell subsets, including IFN-γ⁺CD8⁺, GzmB⁺CD8⁺, CD107a⁺CD8⁺, and perforin⁺CD8⁺, and reduced the presence of severely exhausted PD-1⁺TIM-3⁺CD8⁺ T cells, PD-1⁺LAG-3⁺CD8⁺ T cells and PD-1⁺TIGIT⁺CD8⁺ T cells, thereby transforming immunologically non-inflamed tumors into immune-inflamed tumors.

Conclusions

Our findings suggest that PD-1-directed IL-2 delivered via oVV is both safe and effective, making it a promising candidate for clinical translation.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Many solid tumors are non-inflamed and poorly responsive to immunotherapy.

• Oncolytic viruses can stimulate antitumor immunity, and cytokine-armed viruses enhance this effect.

• Interleukin (IL)-2 has therapeutic potential but causes systemic toxicity.

• Programmed cell death protein-1 (PD-1)/programmed death-ligand 1 (PD-L1) blockade is effective in reinvigorating T cells.

WHAT THIS STUDY ADDS

• A novel oncolytic vaccinia virus expressing a PD-1 ectodomain–IL-2 fusion protein targets IL-2 to tumors via PD-L1 binding.

• This approach enhances local immune activation, reduces systemic toxicity, and converts non-inflamed tumors into inflamed ones.

• Promotes expansion of functional and stem-like CD8⁺ T cells while reducing exhausted subsets.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• Demonstrates a safer, tumor-targeted method for IL-2 delivery.

• Offers a new strategy for converting non-inflamed tumors into immune-inflamed ones.

• Supports further development of multifunctional oncolytic immunotherapies.

Introduction

Cancer immunotherapy has made significant strides in recent years, transforming cancer treatment and offering new hope for many patients, especially those with cancers previously considered difficult to treat. Tumors that are classified as immune-inflamed—characterized by high levels of immune cell infiltration—tend to be more responsive to immunotherapy. Unfortunately, the majority of solid tumors are classified as non-inflamed or immune-desert tumors, which are less amenable to immune-based therapies.¹ As a result, there is an urgent need to develop novel strategies that can enhance T-cell infiltration and shift the cancer-immune equilibrium within the tumor microenvironment (TME),² ultimately transforming non-inflamed tumors into immune-inflamed ones and improving the therapeutic efficacy and broader applicability of cancer immunotherapy.

Oncolytic viruses (OVs) represent a promising new class of cancer immunotherapy agents³ with the potential to transform non-inflamed tumors into more immunologically active ones. OVs can directly infect and kill tumor cells, leading to the release of tumor-associated antigens, neoantigens, damage-associated molecular patterns, pathogen-associated molecular patterns, and inflammatory cytokines. These molecules collectively stimulate both innate and adaptive immune responses, triggering the infiltration of various immune cells, including T lymphocytes, into the TME.^{4 5} OVs have a positive impact on nearly every step of the cancer-immunity cycle. Furthermore, they can be “armed” with chemokines, cytokines, or other bioactive molecules to modulate the TME, thereby enhancing the immune system’s ability to target and treat tumors.6,9

The cytokine interleukin-2 (IL-2) plays a crucial role in the regulation and activation of the immune system, particularly in the expansion and activation of cytotoxic T lymphocytes and natural killer (NK) cells.¹⁰ Historically, IL-2 was one of the first immunotherapies approved by the US Food and Drug Administration (FDA) for cancer treatment, specifically for renal cell carcinoma (RCC) and metastatic melanoma. High-dose IL-2 has shown remarkable responses in a small subset of patients, with durable remissions in some cases. However, its systemic administration is associated with severe adverse effects such as capillary leak syndrome, hypotension, and organ toxicity.^{11 12} Considerable efforts have been devoted to developing IL-2 variants to improve its activity and safety.13,17

Immune checkpoint inhibition has emerged as a groundbreaking strategy in cancer immunotherapy, revolutionizing the treatment of cancers that were previously difficult to manage. Central to this strategy is the targeting of the programmed cell death protein-1 (PD-1)/programmed death-ligand 1 (PD-L1) pathway, which plays a critical role in enabling tumors to evade immune surveillance. PD-1/PD-L1 inhibitors, primarily in the form of monoclonal antibodies (mAbs), have shown significant efficacy in treating various cancers, including melanoma, non-small cell lung cancer, RCC, head and neck squamous cell carcinoma, bladder cancer, and Hodgkin’s lymphoma.¹⁸ To enhance response rates, PD-1/PD-L1 inhibitors are combined with other immunotherapies (such as cytotoxic T-lymphocyte associated protein 4 (CTLA-4) inhibitors,¹⁹ tumor vaccines,²⁰ or OVs²¹) or targeted therapies (such as specific inhibitors for mutant oncogenes).²² Early clinical trials suggest that these combination strategies can boost immune responses and help overcome resistance mechanisms.19,22 OVs engineered to express anti-PD-1/PD-L1 minibodies^{23 24} or the soluble PD-1 (PD-1 ectodomain)²⁵ have shown promising therapeutic effects in preclinical animal models. As an oncolytic virus for cancer therapy, vaccinia virus offers several advantages, including a large transgene capacity, robust intratumoral replication, immune-stimulatory properties, and a well-characterized safety profile in clinical trials.⁵ An oncolytic vaccinia virus vvDD that was modified by inactivating both thymidine kinase (TK) and vaccinia growth factor (VGF) genes has been proven safe in phase 1 clinical trials.26,28 Notably, vvDD treatment in preclinical models was associated with a significant upregulation of PD-L1 expression in the TME.²⁹ IL-2 was engineered into various OVs to investigate the antitumor effects.30,33 Our previous study demonstrated that tumor-localized delivery of membrane-tethered IL-2 via vvDD, in combination with PD-1/PD-L1 blockade, significantly enhanced antitumor immunity by simultaneously promoting T-cell activation and relieving PD-1/PD-L1-mediated immunosuppression.¹⁶ Building on this concept, we hypothesized that a fusion protein consisting of the PD-1 ectodomain and IL-2 (PD-1ecto-R-IL-2) could elicit a potent antitumor effect. In this strategy, the PD-1 ecto portion would mimic the inhibitory function of a PD-L1 antibody while simultaneously facilitating the targeted delivery of IL-2 to a PD-L1-enriched TME, potentially enhancing therapeutic efficacy. In this study, we aimed to investigate whether and how the intratumoral delivery of PD-1ecto-R-IL-2, using the vvDD platform, can generate effective antitumor effects with enhanced safety in mouse tumor models.

Material and methods

Mice and cell lines

Female C57BL/6 (B6 in short) and BALB/c mice, aged 5–6 weeks, were purchased from The Jackson Laboratory (Bar Harbor, Maine, USA) and housed under specific pathogen-free conditions at the Allegheny Health Network Research Institute Preclinical Facility. All animal studies were conducted in accordance with protocols approved by the Institutional Animal Care and Use Committee (IACUC) of the Allegheny Health Network Research Institute under protocol number #1081–2. Normal African green monkey kidney fibroblast CV1 cells, human embryonic kidney 293 (HEK293) cells, human cervical cancer HeLa cells, mouse colon cancer CT26 cells, and Lewis lung carcinoma (LLC) cells were obtained from American Type Culture Collection. Mouse colon cancer cell lines MC38-luc, CT26-luc, and mesothelioma AB12-luc were generated by infecting parental cells with firefly luciferase-carrying lentivirus and selecting with blasticidin. A stable murine PD-L1-overexpressing HEK293 line (HEK293-mPD-L1) was created by transfection with a PD-L1-carrying lentivector and selection with blasticidin. HEK293 cells and HeLa cells were cultured in Dulbecco’s Modified Eagle’s Medium (DMEM) with 20% newborn calf serum, 2 mM L-glutamine, and 1×penicillin/streptomycin, while other cell lines were grown in DMEM with 10% fetal bovine serum (FBS), 2 mM L-glutamine, and 1×penicillin/streptomycin in a 37°C, 5% CO₂ incubator.

Virus generation

vSC20, a vgf gene-inactivated Western Reserve strain vaccinia virus, was a kind from Dr Bernard Moss, NIAID, NIH,³⁴ and used as a parental virus for homologous recombination. The complementary DNA (cDNA) of mouse PD-1 ectodomain (Met1-Gln167) was PCR-amplified from mouse splenic cDNA and inserted into shuttle plasmids pCMS1-IRES-YFP or pCMS1-IRES-FG-YFP,¹⁶ resulting in the creation of new shuttle plasmids, pCMS1-PD-1ecto-IRES-YFP and pCMS1-PD-1ecto-FG-YFP. A DNA fragment containing a rigid linker (EA3K)₄ and mature mouse IL-2 (Ala21-Gln169) was generated based on plasmid pCMS1-IL-2 and subsequently inserted into pCMS1-PD-1ecto-FG-YFP, produce the shuttle plasmid pCMS1-PD-1ecto-R-IL-2-YFP. All primers used for plasmid construction are listed in online supplemental table S1. These shuttle vectors were then used for homologous recombination of PD-1ecto or PD-1ecto-R-IL-2, along with the YFP marker into the tk locus of the vaccinia viral genome of vSC20. To generate the new viruses, CV-1 cells were infected with vSC20 at a multiplicity of infection (MOI) of 0.1 for 2 hours and then transfected with the respective shuttle plasmids. Progeny viruses were harvested 3 days later and used to infect fresh CV1 cells for selection of the new recombinant viruses (online supplemental figure S1). Fluorescence-activated cell sorting was used to select recombinant viruses based on yellow fluorescent protein (YFP) expression, which was assessed 24 hours after infection. As a control, a double gene-inactivated (tk- and vgf-) oncolytic vaccinia virus (oVV) carrying YFP cDNA at its tk locus, referred to as vvDD-YFP (or vvDD), was used in this study.

Transgene expression in vitro

MC38-luc (3.0×10⁵/well), LLC (2.0×10⁵/well), B16 (2.0×10⁵/well), or CT26-luc (2.0×10⁵/well) cells were seeded in 24-well culture plates on day 0. On day 1, the seeded cells were infected with vvDD, vvDD-PD-1ecto, or vvDD-PD-1ecto-R-IL-2 at an MOI of 1 in 0.15 mL of 2% FBS-containing-DMEM per well for 2 hours. After this, 0.35 mL of 10% FBS-containing-DMEM was added into each well, and the cells were cultured for 24 hours. Cell pellets were collected for RNA extraction and subsequent reverse transcription quantitative PCR (RT-qPCR) analysis to measure PD-1 or PD-1ecto-R-IL-2 expression. Cell supernatants from the same culture were harvested to measure IL-2 levels by ELISA. HEK293 (1.0×10⁵/well) or HEK293-mPD-L1 (1.0×10⁵/well) cells were seeded in 48-well culture plates on Day 0. On day 1, the seeded cells were infected with vvDD, vvDD-PD-1ecto, vvDD-IL-2, or vvDD-PD-1ecto-R-IL-2 at MOIs of 0.05, 0.1, 0.5 or 1, respectively. Infected cells were collected 24 hours post-infection for flow cytometry analysis.

Cytotoxicity assay in vitro

Tumor cells were plated at 5×10³ (except MC38-luc cells, which were plated at 1×10⁴) cells per well in 96-well culture plates and infected with indicated viruses the following day at different MOIs. Cell viability was determined 48 hours post-infection using the Cell Counting Kit-8 (Boster Biological Technology, Pleasanton, California, USA).

Rodent tumor models

BALB/c or B6 mice were intraperitoneally (i.p.) inoculated with 4.0×10⁵ AB12-luc or 5.0×10⁵ MC38-luc cancer cells and divided into groups based on tumor size, as determined by live animal IVIS imaging using a Xenogen IVIS 200 Optical In Vivo Imaging System (Caliper Life Sciences, Hopkinton, Massachusetts, USA), respectively. Mice in each group were then i.p. injected with the indicated viruses or phosphate-buffered saline (PBS) on the specified days. In some experiments, mice were euthanized to harvest organs for further analysis.

BALB/c or B6 mice were subcutaneously (s.c.) inoculated with 1.0×10⁶ CT26 or LLC into the right flank. Once tumor volumes reached around 100 mm³, tumor-bearing mice were intratumorally (i.t.) injected with 60 µL of PBS or 60 µL containing 1.0×10⁸ or 5.0×10⁷ plaque forming units (PFU) of virus, diluted with PBS. Tumor volumes were calculated using the standard formula: volume=(length×width²)/2, where length and width were measured using calipers.

Assessment of treatment-related toxicity

Mice treated with the virus were euthanized for collection of blood and lungs 3 days post-treatment. Blood samples were left at room temperature for 2 hours, after which sera were separated by centrifugation. Levels of IL-2 (BioLegend, #431001), interferon (IFN)-γ (BioLegend, #430801), tumor necrosis factor (TNF)-α (BioLegend, #430901), alkaline phosphatase (ALP) (AFG Bioscience: #EK730025), alanine transaminase (ALT) (AFG Bioscience: #EK730024), and aspartate transaminase (AST) (AFG Bioscience: #EK730935) were measured using commercially available kits, following the manufacturers’ instructions. Lung edema was assessed by measuring water content. Briefly, wet lungs were weighed, and then dehydrated at 120°C in a chemical hood until the weight of dry lung stabilized. The difference between the wet and dry lung weights was calculated to determine the extent of edema.

Flow cytometry

Collected tumor tissues were weighed and incubated in Roswell Park Memorial Institute (RPMI) 1640 medium containing 2% FBS, 1 mg/mL collagenase IV (Sigma: #C5138), 0.1 mg/mL hyaluronidase (Sigma: #H6254), and 200 U/mL DNase I (Sigma: #D5025) at 37°C for 1–2 hours to generate single cells. About 200 mg of tumor tissue per mouse, when available, was used to prepare single-cell suspensions. Single cells from tumor tissues or virus-infected cells were blocked with α-CD16/32 antibody (clone 93, eBioscience: #14-0161-85; 1:1000) and then stained with antibodies against mouse CD45 (Brilliant Ultra Violet 395, clone: 30-F11, Fisher Scientific: #36-304-518; 1:300), CD4 (Brilliant Ultraviolet 661, clone: GK1.5, Fisher Scientific: #37-600-4180; 1:300), Foxp3 (Brilliant Violet 421, clone: MF-14, BioLegend: #126419; 1:100), CD8 (Brilliant Violet 650, clone: 53–6.7, BioLegend: #100742; 1:300), PD-1 (Brilliant Violet 605, clone: 29F.1A12, BioLegend; 135220; 1:300), TIM-3 (PE-CF594, clone: 5D12; BD Bioscience: #566998; 1:300), LAG-3 (Brilliant Violet 785, clone: C9B7W; BioLegend: #125219; 1:300), TIGIT (PE/Cyanine7, clone: 1G9; BioLegend: #142108; 1:300), IL-2 (APC, clone: JES6-5H4, BioLegend: #503810; 1:300). The viable cells were assessed using the Zombie Aqua Fixable Viability Kit (BioLegend: #423102). The Cyto-Fast Fix/Perm Buffer Set for intracellular staining was purchased from BioLegend (#426803). Samples were collected on a NovoCyte Penteon Flow Cytometer System (Agilent, Santa Clara, California, USA) and the data were analyzed using FlowJo software (V.10.8.1; TreeStar, Ashland, Oregon, USA).

Reverse transcription quantitative PCR

Total RNA was extracted from virus-infected cells or tumor tissues using the RNeasy Kit (Qiagen, Valencia, California, USA). One microgram of RNA was used for cDNA synthesis, and 25–50 ng of the resulting cDNA was used to conduct messenger RNA (mRNA) expression analysis by TaqMan on the Quantagene q225 qPCR System (Kubo Technology, Beijing, China). All primers for TaqMan analysis were purchased from Thermo Fisher Scientific (Waltham, Massachusetts, USA). Gene expression was normalized to the housekeeping gene HPRT1 and expressed as fold increase (2^−ΔCT), where ΔCT=CT _{(Target gene)} - CT _(HPRT1).

Statistics

Animal survival is shown using Kaplan-Meier survival curves and was statistically analyzed using a log-rank (Mantel-Cox) test (GraphPad Prism V.10.2). Tumor growth curves were statistically analyzed using two-way analysis of variance (ANOVA) (GraphPad Prism V.10.2). Values of p<0.05 were considered statistically significant, and all p values were two-sided. Other statistical analyses were performed using one-way ANOVA with Tukey’s multiple comparison test (GraphPad Prism V.10.2). Data are presented as means±SD. In the figures, standard symbols are used: *p<p0.05; **p<0.01; ***p<0.001; and ****p<0.000; ns indicates not significant.

Data availability

Data are available from the corresponding author on reasonable request.

Results

PD-1ecto-R-IL-2 expressed via oVV tethers on cell membrane

To evaluate the antitumor effect and safety of PD-1ecto-R-IL-2 delivered via vaccinia virus (vvDD), two new viral constructs were developed: vvDD-PD-1ecto and vvDD-PD-1ecto-R-IL-2, designed to express either mouse PD-1ecto or PD-1ecto-R-IL-2 (online supplemental figure S1 and table S1). MC38-luc cells were infected with these viruses or vvDD at an MOI of 1 or treated with PBS. 24 hours post-treatment, total RNA from treated MC38-luc cells was isolated and analyzed by RT-qPCR. The results demonstrated a significant increase in the mRNA levels of both PD-1 and IL-2 in the cells infected with vvDD-PD-1ecto-R-IL-2, while only PD-1 mRNA levels were significantly elevated in cells infected with vvDD-PD-1ecto. The expression of PD-1ecto and PD-1ecto-R-IL-2 was consistent with viral replication, as indicated by similar ratios of PD-1/A34R (a vaccinia virus gene) or IL-2/A34R compared with vvDD infection. Next, supernatants from cells infected with vvDD-PD-1ecto, vvDD-PD-1ecto-R-IL-2, or vvDD at an MOI of 1 were collected 24 hours post-infection to measure IL-2 secretion. As anticipated, IL-2 levels in the supernatant of vvDD-PD-1ecto-R-IL-2-infected cells were significantly higher than those from other oVV-infected cells, indicating successful expression and secretion of PD-1ecto-R-IL-2 (figure 1A). Similar results were observed in three other cancer cell lines (LLC, B16, and CT26-luc) under the same experimental conditions (figure 1B–D). The relative expression of A34R is presented in online supplemental figure S2. Further investigation into IL-2 membrane association was conducted using flow cytometry. To avoid the potential association of murine IL-2 with its receptors on murine cells, we used human HEK293 cells to investigate IL-2 membrane-associated via murine PD-1 ectodomain. This approach was based on the observation that the cross-species binding affinities of murine PD-1 to human PD-L1 and of human PD-1 to murine PD-L1 are similar to the affinities of same-species binding.³⁵ Additionally, we overexpressed murine PD-L1 on HEK293 cells to enhance the IL-2 membrane association; these cells are referred to as HEK293-mPD-L1. HEK293 cells or HEK293-mPD-L1 cells were infected with vvDD, vvDD-IL-2, vvDD-PD-1ecto, vvDD-PD-1ecto-R-IL-2 at MOIs of 0.05, 0.1, 0.5, or 1. Cells were collected 24 hours post-treatment and analyzed by flow cytometry. The results revealed that IL-2⁺ cells were only present in those infected with vvDD-PD-1ecto-R-IL-2, with the percentage of IL-2⁺ cells increasing as the MOI was raised. Notably, the percentage of IL-2⁺HEK293-mPD-L1 cells was significantly higher than that of HEK293 cells at MOIs of 0.1, 0.5 and 1 (figure 2A). Furthermore, the median fluorescence intensity of IL-2 expression on HEK293-mPD-L1 was significantly higher than that on HEK293 cells at an MOI of 1, indicating a greater accumulation of IL-2 on the cell membrane (figure 2B). These results suggested that IL-2 was successfully associated with cell membrane through the interaction between PD-1ecto-R-IL-2 and PD-L1.

Figure 1. PD-1ecto-R-IL-2 is expressed via vvDD. Tumor cells (MC38-luc: 3.0×10⁵ cells (A); LLC: 2.0×10⁵ cells (B), B16: 2.0×10⁵ cells (C) and CT26-luc: 2.0×10⁵ cells (D)) were infected with vvDD, vvDD-PD-1ecto or vvDD-PD-1ecto-R-IL-2 at an MOI of 1. Cell pellets were harvested to measure A34R or transgene expression using RT-qPCR. Tumor cells were treated as mentioned above, and supernatants were collected 24 hours post-infection to quantify the levels of PD-1ecto-R-IL-2 using ELISA. Representative data from one of two independent experiments are shown. IL, interleukin; MOI, multiplicity of infection; PD-1, programmed cell death protein-1; RT-qPCR, reverse transcription quantitative.

Figure 2. PD-1 directs IL-2 to the cell membrane via oVV delivery. HEK293 (1.0×10⁵/well) or HEK293-mPD-L1 (1.0×10⁵/well) cells were seeded in 48-well culture plates on Day 0. On Day 1, the cells were infected with vvDD, vvDD-PD-1ecto, vvDD-IL-2, or vvDD-PD-1ecto-R-IL-2 at MOIs of 0.05, 0.1, 0.5 or 1, respectively. Infected cells were collected 24 hours post-infection and cell surface IL-2 association was analyzed by flow cytometry. Representative flow plots for the highest MOI are shown in (A). The percentages of IL-2⁺ cells are shown in (B, left) and the relative IL-2 MFI is shown in (B, right). MFI values are normalized to the average MFI of the control virus (vvDD) when calculating relative MFI. Data shown are representative of one of two independent experiments. IL, interleukin; MFI, median fluorescence intensity; MOI, multiplicity of infection; oVV, oncolytic vaccinia virus; PD-1, programmed cell death protein-1.

PD-1ecto-R-IL-2 expressed via oVV does not impact viral cytotoxicity

To assess the potential impact of PD-1ecto or PD-1ecto-R-IL-2 expression on viral cytotoxicity in vitro, MC38-luc cells were infected with vvDD-PD-1ecto, vvDD-PD-1ecto-R-IL-2 or vvDD at MOIs of 0, 0.1, 0.5, 1, or 5. Cell viability was measured 48 hours after infection. The results demonstrated that cell viability was similar across all treatment groups, with a dose-dependent decrease in viability observed following infection with vvDD-PD-1ecto, vvDD-PD-1ecto-R-IL-2, or vvDD (online supplemental figure S3A). A similar pattern was observed in the three other cancer cell lines (LLC, B16 and CT26-luc) under the same treatment conditions (online supplemental figure S3B–D). These findings suggest that the expression of either PD-1ecto or PD-1ecto-R-IL-2 does not alter the viral cytotoxicity associated with the PD-1ecto or PD-1ecto-R-IL-2 expressing viruses.

Transgenes are expressed via oVV in the TME

To investigate transgene expressions in vivo, we used a mouse colon cancer MC38-luc model. B6 mice were i.p. injected with MC38-luc cells, and 9 days later, these tumor-bearing mice were i.p. injected with 200 µL PBS or 2.0×10⁸ PFU of vvDD, vvDD-PD-1ecto, or vvDD-PD-1ecto-R-IL-2. The treated mice were euthanized on days 3 and 5 post-treatment, and tumors were collected for RT-qPCR analysis. The results showed no significant difference in A34R expression across all viral treatments, indicating similar levels of viral replication (figure 3A). PD-1 expression was significantly higher in tumors treated with vvDD-PD-1ecto, or vvDD-PD-1ecto-R-IL-2, compared with those treated with vvDD, with no significant difference between the two transgene-expressing treatments (figure 3B,C). IL-2 expression was only significantly elevated in vvDD-PD-1ecto-R-IL-2-treated tumors (figure 3D,E). Compared with PBS treatment, A34R expression was significantly higher, indicating active viral replication. A similar pattern was observed in the tumors collected on day 5 post-treatment (figure 3H–J). Notably, the measured PD-1 and IL-2 levels may include contributions from non-viral sources, as PD-1 is broadly expressed within TME, and tumor-infiltrating T cells may also produce IL-2. These findings demonstrate that viral constructs show comparable viral replication and transgene expression in vivo, consistent with in vitro results.

Figure 3. PD-1ecto-R-IL-2 expression via oVV does not impact viral replication in vivo. B6 mice were i.p. inoculated with 5.0×10⁵ MC38-luc cells and treated with vvDD, vvDD-PD-1ecto, or vvDD-PD-1ecto-R-IL-2 at 2.0×10⁸ PFU/mouse, or PBS, 9 days after tumor cell inoculation, respectively. Tumor-bearing mice were euthanized on days 3 and 5 post-treatment and primary tumors were collected. The expression of A34R (A, D) and transgenes (B, C, E, F) in the TME were analyzed by RT-qPCR. Data shown are combined from two independent experiments for (A–C; n=3 per group). Representative data from one of three independent experiments are shown for (D–F; n=7–12 per group). IL, interleukin; i.p., intraperitoneally; oVV, oncolytic vaccinia virus; PBS, phosphate-buffered saline; PD-1, programmed cell death protein-1; PFU, plaque-forming units; RT-qPCR, reverse transcription quantitative; TME, tumor microenvironment.

PD-1ecto-R-IL-2-expressing vvDD elicited superior antitumor effects with improved safety

To evaluate the antitumor efficacy of vvDD-PD-1ecto-R-IL-2, we used two s.c. tumor models as well as two i.p. tumor models. Initially, we assessed the antitumor activity of vvDD-PD-1ecto-R-IL-2 in the LLC model. B6 mice were s.c. inoculated with 1.0×10⁶ LLC cells into the right flanks and treated when tumors reached around 100 mm³ in tumor volume. Mice were i.t. injected with 60 µL PBS or 5.0×10⁷ PFU of vvDD, vvDD-IL-2, vvDD-PD-1ecto or vvDD-PD-1ecto-R-IL-2. The results showed that treatment with vvDD-IL-2 or vvDD-PD-1ecto-R-IL-2 significantly slowed tumor growth compared with vvDD or vvDD-PD-1ecto by day 24 post-treatment (vvDD-IL-2 vs vvDD, p<0.0001; vvDD-PD-1ecto-R-IL-2 vs vvDD, p<0.0001; vvDD-PD-1ecto-R-IL-2 vs vvDD-PD-1ecto, p<0.0001) (figure 4A). However, there was no significant difference in tumor growth between vvDD-IL-2 and vvDD-PD-1ecto-R-IL-2 treatments (figure 4A). Notably, treatment with vvDD-PD-1 ecto showed results similar to vvDD alone, suggesting that PD-1 ectodomain expression did not provide an additive therapeutic benefit (figure 4A). By day 24, four out of five mice treated with vvDD-IL-2 and three out of five mice treated with vvDD-PD-1ecto-IL-2 were completely cured. On day 24, all mice were euthanized, and tumor-draining lymph nodes (TDLNs) were collected for further analysis. The TDLNs were representative photographed (figure 4B) and weighed (figure 4C). The results showed that the size and weight of tumor-draining lymph nodes from mice treated with either vvDD-IL-2 or vvDD-PD-1ecto-R-IL-2 were significantly smaller than those from other treatments, indicating a reduction in tumor metastasis to the draining lymph nodes. A similar pattern of tumor growth inhibition was observed in the s.c. colon cancer CT26 model (figure 4D). By day 16, two mice each from the PBS-treated, vvDD-treated, and vvDD-PD-1ecto-treated groups developed tumors exceeding 2,000 mm³ and were euthanized in accordance with our IACUC-approved animal protocol. At the same time, we observed significant tumor regression in mice treated with vvDD-IL-2 and vvDD-PD-1ecto-R-IL-2. Specifically, in the vvDD-IL-2 group, three of six mice had tumor volumes below 100 mm³ (14.75, 41.71, and 6.68 mm³), while in the vvDD-PD-1ecto-R-IL-2 group, three of five mice also had tumor volumes below 100 mm³ (52.18, 36.84, and 4.61 mm³)—substantially smaller than the initial average tumor volume at treatment (~100 mm³). We also assessed the antitumor efficacy of vvDD-PD-1ecto-R-IL-2 in i.p. tumor models. BALB/c mice were i.p. inoculated with 4.0×10⁵ AB12-luc cells and treated with 200 µL PBS or 1.0×10⁸ PFU of vvDD, vvDD-IL-2, vvDD-PD-1ecto or vvDD-PD-1ecto-R-IL-2 on day 9 after tumor cell inoculation. The survival data demonstrated that both vvDD-PD-1ecto-R-IL-2 and vvDD-IL-2 treatments significantly extended survival compared with the other treatments, with no significant difference between the two (figure 4E). Interestingly, a few mice died earlier after vvDD-IL-2 treatment, whereas no early deaths were observed in the vvDD-PD-1ecto-R-IL-2 group, suggesting that vvDD-PD-1ecto-R-IL-2 offered superior safety compared with vvDD-IL-2. We have previously shown that vvDD-IL-2 treatment caused severe toxicity in a late-stage MC38-luc peritoneal cancer model.¹⁶ In the current study, we evaluated the effects and safety of vvDD-PD-1ecto-R-IL-2 in the same model. B6 mice were i.p. inoculated with 5.0×10⁵ MC38-luc cells and treated with 200 µL PBS or 2.0×10⁸ PFU of vvDD, vvDD-IL-2, vvDD-PD-1ecto or vvDD-PD-1ecto-R-IL-2 on day 9 after tumor cell inoculation. As expected, vvDD-IL-2 treatment was associated with significant toxicity. However, vvDD-PD-1ecto-R-IL-2 significantly extended survival compared with other treatments, with greatly improved safety (figure 4F). Notably, at the end of the experiments—on days 144 and 153, respectively—the remaining survivors injected with AB12-luc and MC38-luc, who appeared completely healthy, were euthanized to conserve housing resources. On day 5 post-viral treatment, vvDD-IL-2 treatment induced approximately 80% mortality in experimental mice (figure 5A). Serum samples and lung tissues were collected on day 3 post-treatment to evaluate the effects of vvDD-PD-1ecto-R-IL-2 on inflammatory markers. The IL-2 levels in sera from mice treated with vvDD-PD-1ecto-R-IL-2 were significantly lower than those from mice treated with vvDD-IL-2 (figure 5B). We measured the levels of TNF-α, an inflammatory cytokine known to mediate IL-2 toxicity. The results showed that vvDD-IL-2 treatment was associated with significantly elevated TNF-α level in the sera (figure 5C). In contrast, IFN-γ, a cytokine that plays a central role in cytokine release syndrome,³⁶ was significantly reduced in the sera of mice treated with vvDD-PD-1ecto-R-IL-2 compared with those receiving vvDD-IL-2 (figure 5D). Furthermore, mice treated with vvDD-PD-1ecto-R-IL-2 exhibited reduced lung edema (figure 5E) and less liver toxicity, as evidenced by significantly lower serum levels of ALT (figure 5F), and AST (figure 5G), and ALP (figure 5H), compared with the vvDD-IL-2 treated group.

Figure 4. PD-1ecto-R-IL-2-expressing oVV elicits potent antitumor effects. B6 mice (n=4–6 per group) were s.c. inoculated with 1.0×10⁶ LLC cells in the right flank, and these tumor-bearing mice were i.t. treated with 60 µL of PBS or 60 µL of the indicated virus (5.0×10⁷ PFU) per mouse 9 days after tumor cell inoculation. Primary tumor size was measured every 2 days and tumor growth curves are shown (A). All treated mice were euthanized on day 24 post-treatment to collect tumor-draining lymph nodes. Tumor-draining lymph nodes were photographed (B) and weighed (C). BALB/c mice (n=5–6 per group) were s.c. inoculated with 1.0×10⁶ CT26 cells in the right flank, and these tumor-bearing mice were i.t. treated with 60 µL of PBS or 60 µL of the indicated virus (1.0×10⁸ PFU) per mouse, 9 days after tumor cell inoculation. Primary tumor size was measured every 2 days and tumor growth curves are shown (D). BALB/c (n=10–20 per group) or B6 mice (n=19–37 per group) were i.p. inoculated with 4.0×10⁵ AB12-luc or 5.0×10⁵ MC38-luc cells and treated with the indicated viruses at 2.0×10⁸ PFU/mouse or PBS, 9 days after tumor cell inoculation. Kaplan-Meier survival curves are shown (E, F). Data represent the combined results from two independent experiments for (E) and four for (F). IL, interleukin; i.p., intraperitoneally; i.t., intratumorally; oVV, oncolytic vaccinia virus; PBS, phosphate-buffered saline; PD-1, programmed cell death protein-1; PFU, plaque-forming units; s.c., subcutaneously.

Figure 5. PD-1 directed IL-2 via oVV avoids severe toxicity. B6 mice (n=3 per group) were i.p. inoculated with 5.0×10⁵ MC38-luc cells and treated with the indicated viruses at 2.0×10⁸ PFU/mouse or PBS, 9 days after tumor cell inoculation. The treatment-related death rate on day 5 post-treatment is shown (A). Sera and lung tissue were collected 3 days post-treatment. The IL-2 (B), TNF-α (C), IFN-γ (D), ALP (F), ALT (G) and AST (H) levels in the sera were measured. Lung edema is presented as the percentages of water in the lungs (E). Data represent combined results from two independent experiments. ALP, alkaline phosphatase; ALT, alanine transaminase; AST, aspartate transaminase; IFN, interferon; IL, interleukin; i.p., intraperitoneally; oVV, oncolytic vaccinia virus; PBS, phosphate-buffered saline; PD-1, programmed cell death protein-1; PFU, plaque-forming units; TNF, tumor necrosis factor.

PD-1ecto-R-IL-2 delivered by vvDD modulates the tumor microenvironment

To explore how vvDD-PD-1ecto-R-IL-2 elicits antitumor effects, we investigated the TME of MC38-luc tumors post-treatment. B6 mice were i.p. injected with MC38-luc tumor cells, and 9 days later, they received i.p. injections of PBS, vvDD, vvDD-PD-1ecto or vvDD-PD-1ecto-R-IL-2. Treated mice were euthanized, and tumor nodules were harvested on days 3 and 5 post-treatments for RT-qPCR analysis.

The results revealed that antitumor immunity-associated factors, including IFN-γ (figure 6A,E), granzyme B (GzmB) (figure 6B,F) and perforin (figure 6C,G), were significantly upregulated in the tumors following vvDD-PD-1ecto-R-IL-2 treatment on both day 3 and day 5 post-treatment, compared with other viral treatments. Notably, on day 5, the mean relative expression levels of IFN-γ, GzmB and perforin increased from 0.07 to 0.28, 1.29 to 3.76 and 0.29 to 0.76, respectively, compared with day 3, indicating a significant accumulation of these factors following treatment. Interestingly, anti-inflammatory cytokine IL-10 was also significantly upregulated in the tumors after vvDD-PD-1ecto-R-IL-2 treatment on day 5 (figure 6H), compared with the other two viral treatments. As it has been reported that the immunosuppressive cytokine IL-10 expressed by oVV can support prolonged viral persistence, the elevated IL-10 might also facilitate viral replication.³⁷

Figure 6. PD-1ecto-R-IL-2-expressing oVV remodels antitumor effect-associated factors in the TME. B6 mice (n=3 per group for D3 collection; n=4–6 per group for D5 collection) were i.p. inoculated with 5.0×10⁵ MC38-luc cells and treated with the indicated viruses at 2.0×10⁸ PFU/mouse or PBS, 9 days after tumor cell inoculation. Tumor-bearing mice were euthanized on day 3 or 5 post-treatment, and primary tumors were collected. Expression of factors associated with antitumor effects or protumor effects in the TME was analyzed by RT-qPCR. Data represent combined results from two or three independent experiments. CTLA-4, cytotoxic T-lymphocyte associated protein 4; GzmB, granzyme B; IFN, interferon; IL, interleukin; i.p., intraperitoneally; oVV, oncolytic vaccinia virus; PBS, phosphate-buffered saline; PD-1, programmed cell death protein-1; PFU, plaque-forming units; PTGS2, prostaglandin-endoperoxide synthase 2; RT-qPCR, reverse transcription quantitative; TME, tumor microenvironment; VEGF, vascular endothelial growth factor.

We also investigated the expression of immune-suppressive factors, such as vascular endothelial growth factor (VEGF), prostaglandin-endoperoxide synthase 2 (PTGS2), PD-L1 and CTLA-4, within the TME. The results showed that the expression of VEGF and PTGS2 in tumors treated with vvDD-PD-1ecto-R-IL-2 was not significantly different on day 3 post-treatment (figure 6I,J) compared with other viral treatments. However, by day 5, the expression of these factors was significantly downregulated in vvDD-PD-1ecto-R-IL-2-treated tumors compared with those treated with other viral treatments (figure 6M,N). While PD-L1 expression did not show significant differences across all viral treatments (figure 6K,O), the expression of another immune checkpoint, CTLA-4, was significantly upregulated in the tumors treated with vvDD-PD-1ecto-R-IL-2 on both day 3 and day 5, compared with the other viral treatments (figure 6L,P).

Next, we investigated the TME further using flow cytometry. Single cells were prepared from tumor nodules collected on day 5 post-treatment and analyzed. A representative nested gating strategy is presented in online supplemental figure S4. The results showed that the absolute number of CD8⁺ T cells within the tumor mass was significantly higher following treatment with vvDD-PD-1ecto-R-IL-2 compared with other treatment groups (figure 7A). Moreover, functional CD8⁺ T cell subsets, including IFN-γ⁺CD8⁺, GzmB⁺CD8⁺, CD107a⁺CD8⁺, and perforin⁺CD8⁺ T cells, were significantly elevated in the vvDD-PD-1ecto-R-IL-2 group (figure 7B–E). Interestingly, stem-like CD8⁺ T cells (CD8⁺TCF-1⁺) were also increased following vvDD-PD-1ecto-R-IL-2 treatment (figure 7F), indicating enhanced intratumoral T-cell renewal potential and the establishment of a more sustained antitumor immune response. Additionally, vvDD-PD-1ecto-R-IL-2 treatment induced a partial reversal of T-cell exhaustion, evidenced by several key changes. Specifically, the proportions of severely exhausted PD-1⁺TIM-3⁺CD8⁺ T and PD-1⁺TIGIT⁺CD8⁺ T cells within the CD8⁺ T cells were decreased following vvDD-PD-1ecto-R-IL-2 treatment compared with other viral treatments (figure 7G,H). Moreover, the percentage of severely exhausted PD-1⁺LAG-3⁺CD8⁺ T cells (figure 7I) was significantly reduced in the vvDD-PD-1ecto-R-IL-2 group compared with the vvDD-PD-1ecto group. The absolute numbers of regulatory T cells (Tregs, CD4⁺Foxp3⁺) in tumor mass were not significantly different across treatment groups (figure 7J). However, the absolute number of functional CD4⁺Foxp3⁻IFN-γ⁺ T cells was significantly increased following vvDD-PD-1ecto-R-IL-2 treatment compared with other treatments (figure 7K). The ratio of CD8⁺ T cells to Tregs was significantly higher (figure 7L) after vvDD-PD-1ecto-R-IL-2 treatment compared with other treatments, indicating a more favorable immune contexture. Additionally, the proportion of immunosuppressive myeloid-derived suppressor cells (MDSCs, CD11b⁺Gr-1⁺) was decreased (figure 7M) after vvDD-PD-1ecto-R-IL-2 treatment, relative to other viral treatments. Interestingly, expression of the immune checkpoint molecule CTLA-4 was significantly elevated in both CD4⁺ (figure 7N) and CD8⁺ T cells (figure 7O), and PD-L1 expression was markedly upregulated in CD11b⁺ myeloid cells (figure 7P) within the TME after vvDD-PD-1ecto-R-IL-2 treatment. These changes may represent adaptive resistance mechanisms that could dampen the antitumor efficacy of the treatment.

Figure 7. PD-1ecto-R-IL-2-expressing oVV transforms tumor-infiltrating T cells in the TME. B6 mice (n=7–12 per group) were i.p. inoculated with 5.0×10⁵ MC38-luc cells and treated with the indicated viruses at 2.0×10⁸ PFU/mouse or PBS, 9 days after tumor cell inoculation. Tumor-bearing mice were euthanized on day 5 post-treatment, and primary tumors were collected. Tumor-infiltrating cells were analyzed using flow cytometry. Representative data from one of three independent experiments are shown. CTLA-4, cytotoxic T-lymphocyte associated protein 4; GzmB, granzyme B; IFN, interferon; IL, interleukin; i.p., intraperitoneally; LAG-3, Lymphocyte-activation gene 3; oVV, oncolytic vaccinia virus; PBS, phosphate-buffered saline; PD-1, programmed cell death protein-1; PD-L1, programmed death-ligand 1; PFU, plaque-forming units; TCF-1, T cell factor 1; TIGIT, T cell immunoreceptor with Ig and ITIM domains; TIM-3, T cell immunoglobulin and mucin domain-containing molecule 3; TME, tumor microenvironment; Treg, regulatory T cell.

Altogether, vvDD-PD-1ecto-R-IL-2 treatment effectively transformed the TME from an immunologically non-inflamed state into a more inflamed, immune-responsive environment.

Discussion

Immune checkpoint inhibition, particularly targeting the PD-1/PD-L1 pathway, has become a promising cancer immunotherapy, aiming to enhance immune system activity against cancer cells. FDA-approved mAbs, such as cemiplimab, nivolumab, pembrolizumab, atezolizumab, avelumab, and durvalumab, block this pathway to treat various cancers.¹⁸ Small molecule inhibitors are also being explored as alternatives, offering benefits like improved tumor distribution, oral bioavailability, reduced side effects, and lower costs.³⁸ Additionally, soluble PD-1/PD-L1 ectodomains may serve as potential candidates for inhibiting human PD-1/PD-L1, based on the observation that the cross-species binding affinities of murine PD-1 to human PD-L1 and of human PD-1 to murine PD-L1 are similar to the affinities of same-species binding.³⁵ As a proof of concept, it has been reported that soluble PD-1, when used to block the interaction between PD-1 and PD-L1, enhanced vaccine-elicited antigen-specific T-cell responses in mice.³⁹ Moreover, myxoma virus was engineered to express human PD-1 ectodomain (Met1-Thr168), and this recombinant virus demonstrated superior antitumor effects compared with the combination of myxoma virus and anti-PD-1 antibody.²⁵ We investigated the use of PD-1 ectodomain via an oVV, vvDD-PD-1ecto, in cancer treatment. However, this approach did not improve antitumor efficacy over the parent virus, vvDD, in multiple tumor models (figure 4). This may be due to PD-L1 upregulation following oVV infection,²⁹ rather than myxoma infection.²⁵

Cytokines play a crucial role in cancer immunology and immunotherapy, making them promising candidates for cancer treatment. Numerous cytokines have been used to arm OVs, either alone or in combination with other cancer treatments, to enhance antitumor effects.1637 40,48 IL-2 was the first approved immunotherapy for cancer,¹⁰ but it is no longer a frontline treatment due to its severe side effects. We have demonstrated that membrane-bound IL-2, directly delivered into TME by oVV, avoids its severe toxicity while maintaining its antitumor effects in animal models. In combination with PD-1/PD-L1 blockade, this virus cured most of the mice with high tumor burdens.¹⁶ In the current study, we sought to replicate this dual mechanism by engineering a PD-1 ectodomain–IL-2 fusion protein delivered via an oVV. While this construct successfully enhanced IL-2-mediated T-cell activation within the tumor and reduced systemic toxicity (figure 5), it did not fully recapitulate the efficacy of direct PD-1/PD-L1 blockade seen previously (figure 4).¹⁶ We propose that this may be due to insufficient expression levels and duration of the PD-1ecto fusion protein within the TME, limiting its capacity to saturate PD-L1 and prevent its interaction with PD-1 on T cells, especially in the context of increased PD-L1 expression. In contrast to anti-PD-1 antibodies, which can be administered repeatedly to sustain therapeutic concentrations, the efficacy of repeated oncolytic virus dosing may be constrained by the development of antiviral immunity, leading to accelerated viral clearance and a shortened window of transgene expression. Although in vitro assays confirmed binding of the fusion protein to PD-L1, functional blockade in vivo may be incomplete. This limitation highlights the potential benefit of combining vvDD-PD-1ecto-R-IL-2 with systemic anti-PD-L1 antibodies to achieve more effective immune checkpoint blockade. Future studies assessing PD-L1 receptor occupancy and testing such combination therapies are warranted to fully realize the therapeutic potential of this approach. Despite this, the fusion approach helped mitigate the side effects typically elicited by systemic IL-2 administration (figure 5), providing an alternative method for directing IL-2 into TME with reduced toxicity. Although the PD-1 ectodomain may not be the optimal directing vehicle for achieving the dual functions of the fusion protein through the delivery of oVV, this strategy presents a promising approach to minimizing the adverse effects of IL-2 therapy post-optimization of directing vehicle or delivery approaches—such as using engineered T cells (eg, chimeric antigen receptor (CAR)-T cells) or nanoparticle-based platforms.

The PD-1ecto-R-IL-2-expressing oVV also preserved the antitumor function of IL-2 compared with other viral treatments and elicited potent antitumor effects in all the four tested tumor models (figure 4). Treatment with vvDD-PD-1ecto-R-IL-2 increased the expression of the antitumor-related factors, including IFN-γ, GzmB and perforin, while reducing protumor factors such as VEGF and PTGS2 (figure 6A–N). VEGF is a key factor in the development and growth of cancer, involved in many events in cancer such as tumor angiogenesis, tumor cell signaling, abnormal blood vessels formation and cancer cell progression.⁴⁹ IL-2 can both increase and decrease VEGF levels, depending on the conditions and the type of immune cells involved.^{50 51} It has been reported that the treatment with low-dose IL-2 in combination with zoledronic acid increased serum VEGF levels in patients with cancer.⁵⁰ However, in the vvDD-PD-1ecto-R-IL-2 treatment, IL-2 accumulated in the TME via PD-1 targeting to PD-L1, which is abundant in the TME, especially after oVV treatment.²⁹ The high concentration of IL-2 may decrease VEGF expression. PTGS2 encodes cyclooxygenase 2, whose primary downstream product, prostaglandin E2 (PGE₂), is a known negative regulator of immune response in the TME.⁵² PGE₂ has been shown to inhibit IL-2 sensing in tumor-infiltrating lymphocytes by downregulating the expression of IL-2Rγ_c chain, which disrupts the assembly of IL-2Rβ-IL-2Rγ_c membrane dimers.⁵³ The high concentration of IL-2 accumulated following vvDD-PD-1ecto-R-IL-2 treatment may overcome the inhibitory effect of PGE₂ through a positive feedback loop: as IL-2 levels increase, receptor expression is upregulated, which in turn further downregulates PGE₂ expression.

There was no significant difference in PD-L1 expression in the three viral treatments (figure 6K,O). However, a significant increase in intratumoral CTLA-4 expression was observed following vvDD-PD-1ecto-R-IL-2 treatment compared with the other two viral treatments (figure 6L,P). This increase in CTLA-4 expression may attenuate antitumor effects, but it also suggests the potential for combination therapy with anti-CTLA-4 antibodies to further enhance antitumor effects.

In addition to elevated levels of antitumor factors and reduced levels of protumor factors, we observed an increase in the absolute number of tumor-infiltrating CD8⁺ T cells (figure 7A), including functional IFN-γ⁺CD8⁺, GzmB⁺CD8⁺, CD107a⁺CD8⁺, and perforin⁺CD8⁺ T cells (figure 7B–E). Interestingly, stem-like CD8⁺ T cells (CD8⁺TCF-1⁺) were also increased following vvDD-PD-1ecto-R-IL-2 treatment (figure 7F), which is consistent with the observed reduction in PTGS2 levels (figure 6N), as the PTGS2 downstream product PGE₂ has been reported to limit the expansion of tumor-infiltrating stem-like CD8⁺ T cells.⁵⁴ These findings suggest enhanced intratumoral T-cell renewal capacity and the potential for a more durable antitumor immune response. Meanwhile, the proportions of exhausted PD-1⁺TIM-3⁺CD8⁺ T cells (figure 7G), PD-1⁺TIGIT⁺CD8⁺ T cells (figure 7H), and PD-1⁺LAG-3⁺CD8⁺ T cells (figure 7I) were decreased following vvDD-PD-1ecto-R-IL-2 treatment. The co-expression of PD-1 and TIM-3,⁵⁵ LAG-3,⁵⁶ or TIGIT⁵⁷ on T cells is associated with more severe exhaustion. Notably, vvDD-PD-1ecto-R-IL-2 treatment altered these co-expression patterns on tumor-infiltrating CD8⁺ T cells, suggesting potential mechanisms by which it exerts its antitumor effects.

Recent studies have highlighted that critical populations of stem-like CD8⁺ T cells are essential for effective responses to immune checkpoint blockade and are often localized not only within tumors but also in TDLNs, the spleen, and tertiary lymphoid structures.^{58 59} These peripheral and lymphoid-resident pools serve as reservoirs for T-cell renewal, playing a vital role in sustaining long-term antitumor immunity. In our study, while we observed a significant increase in intratumoral CD8⁺TCF-1⁺ cells following vvDD-PD-1ecto-R-IL-2 treatment, the intratumoral delivery of the PD-1-IL-2 fusion protein may limit its engagement with stem-like CD8⁺ T cells residing in peripheral lymphoid compartments. This spatial constraint could also contribute to the incomplete blockade of the PD-1/PD-L1 axis and the less robust therapeutic efficacy compared with systemic checkpoint blockade strategies. These considerations offer a potential explanation for the observed limitations and underscore the need for future studies evaluating systemic or combination approaches to effectively target stem-like CD8⁺ T cells across relevant anatomical sites.

NK cells play a multifaceted role in oncolytic virotherapy, either supporting or hindering the effectiveness of treatment.⁶⁰ In the case of vaccinia virus, MDSCs rapidly accumulate at the site of VV infection, where they negatively regulate NK cell activation and function in response to the virus.⁶¹ In a colon cancer model, there was no significant difference in the potent antitumor effects elicited by membrane-bound IL-2 expressing oVV in the presence or absence of NK cells. Interestingly, a slight survival benefit was observed in the absence of NK cells.¹⁶ These results suggest that NK cells may play a redundant role in the induction of antitumor effects with IL-2-expressing oVV, despite IL-2 being known to enhance the activity of both NK cells and T cells.¹⁰ That is the reason why we did not investigate NK cell function in the current study.

In summary, our study demonstrates that PD-1-directed IL-2 into the TME via oncolytic oVV enhances antitumor immune factor levels and increases tumor-infiltrating CD8⁺ T cells, while reducing protumor factors and severely exhausted T cells. This approach effectively transforms immunologically non-inflamed tumors into immunologically immune-inflamed tumors, eliciting potent antitumor effects while avoiding the severe toxicity associated with systemic IL-2 administration. Additionally, our findings suggest the potential for combination therapy with anti-CTLA-4 antibody to enhance antitumor effects in the future. In conclusion, PD-1-directed IL-2 delivered via oVV demonstrated both safety and efficacy, offering strong potential as a candidate for clinical translation.

Data availability statement

Data are available upon reasonable request.