Introduction

Oncolytic viruses (OV) are pleiotropic modalities for cancer immunotherapy. Their initial mechanism was based on their specific properties to infect and preferentially replicate in tumor cells, inducing their direct lysis. It is now scientifically well acknowledged that their antitumoral activity is mostly mediated by immune mechanisms, including production of type I interferon, recruitment and activation of antigen-presenting cells (APC) and T cells, tumor antigens uptake and presentation by APC, and priming of tumor-specific T cells.1 Despite multiple successful clinical trials, the commercial development of OVs is still rather modest.2 Thus, preclinical research and innovative OV designs remain to be of high need. In this perspective, the ability to modify OV’s genomes is largely exploited, enabling the expression of a large diversity of recombinant transgenes (cytokines, membrane receptor ligands, enzymes, transporters, antibodies, etc) at the site of virus replication, resulting in the changes in the tumor microenvironment (TME).2 3 The tumor-restricted replication of OV allows vectorizing potent molecules with narrow therapeutic windows, for which both a high intratumor concentration and a low blood concentration are required.4 Such features are those of both interleukin-12 (IL-12) and anti-cytotoxic T-lymphocyte-associated antigen-4 (@CTLA-4) monoclonal antibodies, for which tolerability issues have limited clinical use.5 6

The mechanisms of these very potent immune effectors were largely documented.

IL-12 induces proliferation and activates natural killer (NK) and T cells (particularly CD4+Th1 T cells) which in turn secrete interferon-gamma (IFN-γ). The secreted IFN-γ is the probable cause of the poor tolerability of IL-12.5 IFN-γ displays broad antitumor activity; direct inhibition of tumor cell growth, activation of T and NK cells, skewing of macrophages towards M1 phenotype, and activation of dendritic cells. However, antitumor properties of IFN-γ are counter-balanced by several immunosuppressive feedbacks such as the upregulation of PD-L1 or recruitment of regulatory T cells (Treg).7

@CTLA-4 antibodies were reported to act via several non-exclusive mechanisms, some of which are still being debated.8–10 Despite proven antitumoral activities, alone or in combination with other cancer treatment modalities in several indications, their clinical use remains hindered by poor tolerance, leading to treatment discontinuation in up to 19% of patients.6

To enhance their on-target, and/or limit the off-tumor toxicities, IL-12 or @CTLA-4 have been vectorized in viral vectors, oncolytic or not,11–14 as well as DNA15 or RNA16 vectors, mostly for intratumoral administration.17 A combination of OV-vectorized IL-12 and @CTLA-4 administered systematically demonstrated significant preclinical activity in glioma.18 The vectorization of IL-12 and @CTLA-4 in an oncolytic Herpes Simplex Virus (HSV) was also reported but with three other immunomodulatory transgenes, making it difficult to specifically attribute the biological activity to each single component.19

Vaccinia virus (VACV) is particularly interesting as a platform for IL-12 and @CTLA-4 vectorization due to its high capacity of genome insertion (up to 25 kb), safety track record, and ability to be administered intravenously (IV).20

We report here the generation and characterization of TG6050, a new VACV encoding both IL-12 and @CTLA-4 antibody. The infection of tumor cells with TG6050 induces the expression at high levels of functional IL-12 and @CTLA-4 in vitro and in vivo. The three components of TG6050: (oncolytic viral backbone, IL-12 and @CTLA-4) acted together to induce a tumor regression in numerous murine tumor models. Moreover, the use of different and complementary methods demonstrated that TG6050 triggers a strong adaptive antitumoral immune response, along with a profound modification of the TME immune landscape. Hence, the combination of TG6050 with anti-programmed cell death protein-1 (PD-1) resulted in improved antitumoral activity in murine tumor models. Our toxicology study showed that TG6050, when administered IV to cynomolgus monkeys, was well tolerated, with circulating IL-12 concentrations remaining below toxic levels.

Materials and methodsViruses

All viruses constructed and used in this publication were VACV Copenhagen strain deleted of genes coding for: i)thymidine kinase (TK or J2R), ii)the large subunit of ribonucleotide reductase (RR or I4L) and iii) M2L genes except TG6002 (VV benchmark)20 which is only TK-RR-. Viruses were generated by homologous recombination in chicken embryo fibroblast (CEF). CEF were transfected with the shuttle plasmids encoding the expression cassettes and previously infected with parental virus encoding green fluorescent protein (GFP) or mCherry at each locus of transgene insertion (i.e., J2R and I4L). Recombinant viruses were isolated by picking non-fluorescent viral plaques. The identity and integrity of expression cassettes were checked by DNA sequencing of PCR amplicons of each locus. TG6050 encodes human single-chain IL-12 and a whole monoclonal IgG1 antibody (mAb) targeting human CTLA-4. mTG6050 encodes a murine single-chain interleukin 12 (mIL-12) and a whole murine mAb anti-murine CTLA-4 (m@CTLA-4) IgG2a (9D9). VV mIL-12 encodes mIL-12. VV m@CTLA-4 encodes 9D9. VV CTRL either does not encode any transgene (VV) or encodes luciferase enzyme at the TK locus (VV-Luc), see each figure legend for specification. All VACVs are produced on CEF as described previously21 and titrated on Vero cells (see online supplemental material for details).

Cell lines

Murine tumor cells CT26, B16F10, Renca, and human tumor cells HeLa, A549, HT-29, MIA PaCa-2, SK-MEL-1, MDA-MB-231 cell lines were purchased from American Type Culture Collection (ATCC) and grown according to ATCC protocols. The human head and neck squamous cell carcinoma cell line CAL33 was isolated in the Centre Antoine-Lacassagne (Nice, France).22 Lewis lung carcinoma 1 (LLC1) and EMT6 cell lines were properties of and used by Oncodesign.

Statistical analysis

In vivo statistical analysis: Survival analysis was performed with Kaplan-Meier curves and groups were compared with a log-rank test or with a Cox model. A pairwise comparison with the Tukey method for multiplicity adjustment was then performed between the different group levels. The tumor volume was converted to its equivalent diameter with the following formula: Dequivalent= ∛(6V⁄π)23 and the latter was considered as representative of the tumor size. Tumor growth over time was analyzed using a quadratic mixed model. A post hoc pairwise comparison of the tumor diameter mean for the different groups was then performed using Tukey’s adjustment.

ELISpot analysis: General linear models were used to evaluate differences in the mean spot forming units between treatment groups on day 21. The model was built with an interaction term between the group and the day. Pairwise comparisons between groups were performed using Tukey’s correction.

Flow cytometry analysis: the non-parametric Kruskal-Wallis test was used to test for the overall experiment effect. In order to provide specific information about which groups were affected, post hoc comparisons using Wilcoxon’s test and Dwass-Steel-Critchlow-Fligner adjustment were conducted.

Cytokine analysis: A multiple linear regression model with a log-transformed dependent variable (cytokine concentrations) was used. An interaction term between two categorical variables; time points and group, was constructed to examine the differences between groups and whether this effect varies across time.

Analyses were conducted using SAS V.9.4 and R V.4.3.1. The level of significance was set to 5%.

Animal experiments

Animal experiments were approved by ethical committee: "Comité d'éthique en matière d'expérimentation animale de Strasbourg (CREMEAS, CEEA - 035)".

Other materials and methods are described in detail in online supplemental file.

ResultsTG6050 retains its viral features while allowing expression of functional IL-12 and anti-CTLA-4.

Transgenes encoding full-length @CTLA-4 IgG1 human antibody and single-chain IL-12 (p40 fused to p35 subunit24) were inserted into Copenhagen VACV genome at J2R and I4L loci and downstream signal peptides to ensure their secretion by infected cells (figure 1A).20 The same viral early/late promoter, pH5R, was used to control the independent heavy and light chain expression, optimizing the assembly of a full-length antibody (online supplemental figure S1). The transcription of IL-12 transgene is controlled by the late pF17R promoter, minimizing its expression in non-permissive normal cells. Murine surrogate viruses, including mono-armed ones, were also constructed by inserting transgenes encoding single-chain mIL-12 and/or murine IgG2a @CTLA-4 (9D9) (figure 1A). In addition to the J2R and I4L deletions, the virus genome carries a deletion of the M2L gene (online supplemental figure S2). Recent studies, from our group and others, showed that the viral protein M2 binds CD80 and CD86, blocking their interactions with CD28 and CTLA-4, and consequently inhibiting the priming of T cells.21 25 26 The activity of an @CTLA-4 was abrogated in the culture supernatants of M2L+ VACV-infected A549 and restored with M2L– variant (online supplemental figure S3). These results indicated that the level of secreted M2 protein was sufficient to block the CD80/86 co-stimulatory signals under current infection conditions. This prompted us to engineer TG6050 from a viral backbone deleted for the non-essential M2L gene. The oncolytic activity and the replication of TG6050 were assessed in six human tumor cell lines. Additionally, replication in the production cells (HeLa and CEF) and healthy cells was also assessed. Both replication and oncolytic activity were found to be comparable to those of the reference double deleted vector, TG6002 (figure 1B,C, online supplemental figure S4-6, see also oncolytic activities of mTG6050 online supplemental figure S7). The expression of both transgenes in culture supernatants of infected cells was quantified by ELISA. @CTLA-4 and IL-12 were both expressed in all tested tumor cell lines, with levels ranging from 0.5 to 4.2 µg/mL for @CTLA-4 and 0.2 to 6.4 µg/mL for IL-12 figure 1D (see also expression of murine transgenes online supplemental figure S8). The biological activity of IL-12 produced by the six infected tumor cells was similar to a reference recombinant IL-12 protein (figure 1E). The checkpoint blockade (figure 1F) and antibody dependent cell cytotoxicity (ADCC, (figure 1H) activities of TG6050-produced @CTLA-4 were assessed using two cell-based immune assays and showed biosimilarity to a reference product. N-linked glycosylation of different batches of purified @CTLA-4 (online supplemental figure S9) was investigated, as these post-translational modifications are variable in tumor cells27 and crucial for ADCC activity.28 The results revealed substantial differences in the glycoprofiles of @CTLA-4 produced by infected tumor cells, and differences from the reference antibody: produced in Chinese Hamster Ovary (CHO) cells figure 1G (online supplemental figure S10). None of the analyzed samples contained defucosylated antibodies known for increased affinity to Fc receptors and thus higher ADCC.29 Surprisingly, despite the large glycosylation heterogeneities, the diverse @CTLA-4 variants produced by human tumor cells, displayed similar, or higher activities than those measured for reference @CTLA-4 (figure 1H). These findings collectively underscore the capability of TG6050 to induce expression of high levels of functional @CTLA-4 and IL-12, while preserving the functional oncolytic characteristics of the VACV vector.

Figure 1

TG6050 features and in vitro characterization. (A) Schematic representations of recombinant vaccinia virus used for this work; single-chain IL-12 under pF17R and light chain @CTLA-4 under pH5R viral promoters were inserted at the I4L locus. Heavy chain @CTLA-4 under pH5R or luciferase under p11k7.5 viral promoters were inserted at the J2R locus. The M2L-deleted locus of all viruses, except TG6002, is not represented. (B) Oncolytic activities of TG6050 and benchmark viruses (VV-Luc (VV CTRL) and TG6002), represented as EC50 (MOI)±SEM in six human tumor cell lines. (C) Replication of TG6050, VV-Luc (VV CTRL) and TG6002 after 72 hours in six human tumor cell lines (left graph), and replication of TG6050, VV wild type (WT, without any genome deletion) and TG6002 in normal human hepatocytes. Titers are in total PFU/sample. (D) Expression levels of @CTLA-4 and IL-12 in the culture medium of six human tumor cell lines. Cells were infected at MOI 0.01 with TG6050 for 3 days. IL-12 and @CTLA-4 were quantified by ELISA. Results are represented as means (μg/mL)±SD. (E) IL-12 biological activity in culture media of six TG6050-infected cells was determined by the HEK-Blue assay. The enzymatic activity of the reporter protein secreted on engagement of IL-12R is measured by absorbance at 630 nm. Recombinant IL-12 was used as a benchmark. (F) @CTLA-4 ICI activity in 20-fold concentrated culture media of six TG6050-infected cells. Luciferase enzymatic activity was measured by luminescence. Recombinant @CTLA-4 (CHO) was used as a benchmark. (G) N-glycosylation profiles of @CTLA-4 were produced and purified from the culture media of the six TG6050-infected cells. Results are represented as % of total glycoforms and compared with the benchmark @CTLA-4 (CHO). (H) @CTLA-4 ADCC activity in 20-fold concentrated culture media of six TG6050-infected cells is assessed using effector Jurkat-CD16 and target Raji-hCTLA-4 cells. Luciferase enzymatic activity was measured by luminescence. Results are represented as % of the signal at the highest concentration of benchmark @CTLA-4 (CHO). @CTLA-4, anti-cytotoxic T-lymphocyte-associated antigen-4; IL, interleukin. CHO Chinese Hamster Ovary cells, ICI immune checkpoint inhibitor, MOI multiplicity of infection, and PFU plaque forming unit.

Intratumoral injection of TG6050 increased pro-inflammatory cytokines in tumor

In order to monitor the transgenes’ expression in vivo, mTG6050, a surrogate murine product (figure 1A) was injected intratumorally (IT) three times (D7, 9, 11 after tumor cell implantation) at a dose of 1E+07 plaque forming unit (PFU) in mice bearing B16F10 tumors. On D7, 9, 14, and 18, the tumor and the blood were collected for virus titration and measurement of 20 pro-inflammatory cytokines using Luminex (figure 2A). The murine @CTLA-4 IgG2a 9D9 could be quantified only in tumor samples because of interference with endogenous IgG in serum. 9D9 was accumulated at high levels, and these levels remained stable in the tumor throughout the experiment (figure 2B). Similarly, the concentrations of mIL-12 in tumors of mice injected with mTG6050 were up to 10,000-fold higher than those in the control groups (figure 2B). The circulating concentrations of mIL-12 increased transiently by approximately 1,000-fold in mice treated by mTG6050, but remained at safe levels, as suggested by the absence of weight loss (figure 2C, online supplemental figure S11). Since IL-12 induces the secretion of IFN-γ by T and NK cells, its concentration in both tumor and serum also increased (figure 2B,C). Several inflammatory cytokines, including CCL5, IL-18, CXCL10, IL-2, CCL3, GM-CSF, TNFα, CCL4, CCL11, CCL2, CXCL1, CXCL2, IL-4, IL-6, IL-5, IL-13, IL-1β and CCL7, were significantly increased in the tumor after treatment with the control vector VV CTRL, for at least one time point. Moreover, the levels of IL-18, IL-2, GM-CSF, TNFα, IL-5, IL-13, CXCL1 and IL-1β were further significantly increased by treatment with mTG6050 (online supplemental figure S12). Interestingly, virus titer in the tumor remained stable during the 11 days of the experiment and at similar levels to those of the control virus, indicating that IFN-γ did not inhibit VACV replication (figure 2B). No viral particles were found in the blood of mice at any time point, indicating a rapid clearance of the virus outside the tumor.

Figure 2

Pharmacokinetics of virus, mIL-12 and IFN-γ after intratumoral injections. (A) Schematic representation of the experimental design: C57BL/6 mice bearing B16F10 tumors were treated (or not) intratumorally three times at D7, D9 and D11 with either VV-Luc (VV CTRL) or mTG6050 at 1E+07 PFU/injection. Subsequently, three mice/group/time point were euthanized for tumor, blood, and serum collection at D9 before the second treatment, D11 before third treatment, D14 and D18. (B–C) Virus was titrated on Vero cells, 9D9 was measured by ELISA and mIL-12, mIFN-γ (and 18 other pro-inflammatory cytokines) were measured by Luminex in: (B) tumors, and in (C) blood (virus) or sera (cytokines) of mice. No virus particles were detected in the blood of treated mice. IT, intratumorally; mIL, murine interleukin; mIFN, murine interferon; and SC, subcutaneous.

mTG6050 displayed antitumoral activity in “hot” and “cold” tumors and induced long-term T-cell memory response

The antitumoral activity of mTG6050, and the contribution of each functional component of the product were assessed IT or IV on both immunosensitive (“hot”; CT26, EMT6) and immunoresistant (“cold”; B16F10) syngeneic murine tumors implanted subcutaneously (figure 3A). The CT26 “two tumors” model, where only one flank is treated IT, was implemented to observe any potential distant responses (abscopal effect, figure 3A). Intravenous administration of mTG6050, but not VV CTRL, showed a significant antitumoral activity in CT26 (single tumor, figure 3B) and in B16F10 (online supplemental figure S13). mTG6050 was further injected IT to compensate the weaker replication of VACV in murine tumor cells compared to human tumor cells. For CT26 and EMT6, the lowest active administered dose was 1E+05 PFU (online supplemental figure S14 and S15) and the maximal effect was observed at doses of 1E+06 and 1E+07 PFU, resulting in 60–100% of tumor-free mice, respectively (online supplemental figure S14 and S15). In the CT26 model, surviving mice were rechallenged at 123 days after the first tumor implantation. Most of the animals (6/8) that survived the initial challenge did not develop any tumors when rechallenged with the same CT26 cells (five times the initial amount), but they were not protected against grafting with Renca cells (figure 3C). These results demonstrate that treatment with mTG6050 generated a specific and long-lasting antitumoral immunity.

Figure 3

Antitumoral activity of mTG6050 in syngeneic murine tumor models. (A) Schematic representation of experimental designs: mice bearing either B16F10 (n=10) or CT26 (n=10) tumor on one flank (“one tumor”), or “two tumors” CT26 (both flanks, n=10) were treated (or not) intratumorally (IT) or intravenously (for CT26 “one tumor” model only) three times at D7, D9 and D11 post cell implantation with either VV CTRL, VV mIL-12, VV m@CTLA-4 or mTG6050 at 1E+07 PFU/injection. In the CT26 “two tumors” experiment, three mice/group were euthanized at D15, and both injected and non-injected tumors were collected for virus titration. Tumor volume (mm3) and animals’ weight (g) were assessed twice a week for up to 90 days. (B) Mice bearing CT26 tumors (n=10) were treated intravenous (or not) as described above with VV-Luc (VV CTRL) or mTG6050. Graphs show the mean tumor volume (mm3, left panel) and percentage of survival (right panel). For tumor growth, statistically significant differences (S, with adjusted p<0.01 by pairwise comparison using Tukey’s adjustment) were observed starting from D15 or D18 compared with non-treated or VV CTRL, respectively. For survival analysis, statistically significant differences (****p<0.0001 by log-rank test) were observed compared with non-treated group. (C) Mice that rejected CT26 (single) tumors after IT treatment by mTG6050 (see online supplemental figure S14) were challenged with fivefold more CT26 or Renca (control) tumor cells 123 days after first cell implantation. Graphs show mean tumor volume (mm3) (left panel) and percentage of survival (n=8 and n=7 for CT26 and Renca rechallenge, respectively, n=5 for naïve mice challenged with CT26 and Renca tumors, right panel). For tumor growth, statistically significant differences (S, significant with adjusted p<0.05 by pairwise comparison using Tukey’s adjustment) were observed after rechallenge starting from D4 or D18 compared with CT26 control or Renca rechallenge, respectively. For survival analysis, statistically significant differences (**p<0.01 by log-rank test) were observed compared with CT26 control group. (D) Mice bearing B16F10 tumors (n=10) were treated IT as described in (A). Graphs show mean tumor volume (mm3) (left panel) and percentage of survival (right panel). For tumor growth, statistically significant differences (S, with adjusted p<0.05 by pairwise comparison using Tukey’s adjustment) were observed starting from D9 for both mTG6050 and VV mIL-12 and from D11 for VV CTRL and VV m@CTLA-4, compared with non-treated group. For survival analysis, statistically significant differences (*p<0.05 and ****p<0.0001 by log-rank test) were observed compared with non-treated group. (E) Mice bearing “two tumors” CT26 (n=10) were treated IT as described in (A). Graphs show mean tumor volume (mm3) of the treated and non-treated (contralateral) tumors (left panel) and percentage of survival (right panel). For tumor growth, statistically significant differences (s/S for contralateral tumor and treated tumor, respectively) with adjusted p<0.05 by pairwise comparison using Tukey’s adjustment were observed starting from D14, D17 or D21 compared with non-treated. For survival analysis, statistically significant differences (*p<0.05 and ****p<0.0001 by log-rank test) were observed compared with non-treated group. IT, intratumorally; m@CTLA-4, murine anti-cytotoxic T-lymphocyte-associated antigen-4; mIL, murine interleukin ; PFU plaque forming unit and SC subcutaneous.

In the cold B16F10 tumor model, intratumoral mTG6050 treatment significantly delayed tumor growth without resulting in its complete eradication (figure 3D). IL-12 contributed mainly to the antitumoral activity in this model, whereas both IL-12 and @CTLA-4 were necessary for the optimal survival in CT26 “two tumors” model, (figure 3D,E). It is noteworthy that the abscopal effect in the CT26 model was not due to diffusion of mTG6050 to the non-injected tumor, as the latter was free of viral particles (online supplemental figure S16). Finally, all these antitumoral activities were observed without the typical weight loss associated with IL-12 systemic treatments (online supplemental figure S17).

mTG6050 treatment activated IFN-γ pathway and induced massive infiltration of TME by innate and adaptive immune cells

Selective depletion of CD8+, CD4+T and NK cells was performed to identify the immune mechanisms involved in the response to mTG6050 in the CT26 model. The antitumor activity of mTG6050 was almost completely abolished after depletion of CD8+T cells, while the depletion of CD4+T cells partially inhibited TG6050’s antitumoral activity (figure 4A). Depletion of NK had minimal, if any effect on tumor, however, their involvement could not be completely excluded, as their depletion was transient (online supplemental figure S18).

Figure 4

mTG6050 treatment triggers IFN-γ secretion and pro-inflammatory pathways. (A) Mice bearing CT26 tumors were injected IP with CD4+ or CD8+ T (n=10) or NK cells (n=15) depleting antibody or their respective isotype controls (n=10 for @ T-cell isotype control and n=15 for @NK isotype control) at D-3 and D4, then were treated (or not) intratumorally three times at D7, D9 and D11 with mTG6050 at 1E+07 PFU/injection. Graphs show mean tumor volume (mm3) (left panel) and percentage of survival (right panel). For tumor growth, statistically significant differences (S, with adjusted p<0.01 by pairwise comparison using Tukey’s adjustment) were observed starting from D11 or D18 compared with mTG6050+@CD4 or mTG6050+@CD8, respectively. For survival analysis, statistically significant differences (***p<0.001 by log-rank test) were observed compared with mTG6050+@CD8 group. (B) Mice bearing either CT26 or B16F10 tumors were treated (or not) intratumorally three times at D7, D9 and D11 with buffer (n=5), 1E+07 PFU/injection VV-Luc (VV CTRL, n=10) or mTG6050 (n=10). Spleens from CT26 bearing mice were harvested at D21 and an ELISpot IFN-γ was performed. In another experiment, at D11, before the third treatment, and at D17/18, CT26 or B16F10 tumors (n=5) were harvested, and infiltrating immune cells were assessed by flow cytometry. (C) Results represent the mean number of spot-forming units per 1E+06 splenocytes after restimulation with AH1 CT26 peptide, VACV peptide or whole killed CT26 cells. Irrelevant peptide was used as control. *Adjusted p<0.05 and ****adjusted p<0.0001 by pairwise comparison using Tukey’s adjustment compared with VV CTRL group. (D) Proportion of CD45, CD8, Treg, M1 and M2 macrophages and PD-L1 positive cells infiltrating CT26 (upper panels) or B16F10 (lower panels) tumors at D11 after treatment (or not) with VV (VV CTRL) or mTG6050. NS=not significant, *adjusted p<0.05 by pairwise comparison using Dwass-Steel-Critchlow-Fligner adjustment. IFN, interferon; IP, intraperitoneal; NK, natural killer; PD-L1, programmed death-ligand 1; PFU, plaque forming unit; SC, subcutaneous;Treg, regulatory T cells.

To further investigate the immune response triggered by mTG6050, spleens of mice bearing CT26 tumors were harvested 14 days post-first intratumoral treatment (figure 4B). The treatment with mTG6050 significantly increased the number of tumor-specific IFN-γ secreting lymphocytes (ie, sensitive to restimulation by AH1 peptide, or killed CT26 cells) compared with VV CTRL treatment (figure 4C). Interestingly, the immune response against a viral vector antigen did not increase to the same extent.

Immune infiltration was characterized by flow cytometry in both CT26 and B16F10 models (figure 4B). Although the small number of samples did not allow us to show statistically significant differences for some populations, general trends were observed. As soon as 4 days after the first injection, a non-significant increase in the infiltration of CD8+ and a statistically significant decrease of Treg (figure 4D) were observed in CT26 model. Except on CT26 cells at day 17, programmed death-ligand 1 (PD-L1) marker increased significantly for the two models and on both immune and tumor cells after mTG6050 treatment versus non-treated and/or VV CTRL groups (figure 4D, online supplemental figure S19). In B16F10 tumors, activated CD8+KLRG1+ T cells significantly increased after treatment with mTG6050 compared to either the VV CTRL or non-treated groups (online supplemental figure S20). Moreover, the ratio of M1 macrophages over M2 also showed a trend for an increase at the same time point in both models (figure 4D). Same analyses performed later (8 or 9 days post-first administration) confirmed the Treg depletion in both models, and PD-L1 increase but only for B16F10 model (online supplemental figure S19 and S21).

These results were confirmed by transcriptomic studies in which CT26 and B16F10 tumors were treated and harvested as previously described. Both tumors were analyzed for differential expression of total messenger RNAs by 3’RNA sequencing (figure 5A), while CT26 tumor was also analyzed additionally by single cell RNA (scRNA) sequencing (figure 5B). Treatment with mTG6050 induced drastic modification of tumor transcriptomes (online supplemental figure S22). A gene set variation analysis using signature markers of immune cell populations and pathways30 showed a strong increase of most immune pathways, such as CD8+T cells, NK cells and the type II IFN response (figure 5A). Notably, the earlier time point (4 days after first virus injection) in the CT26 model showed strong activation of type I IFN, cytolytic activity, and major histocompatibility complex class I pathways but a moderate induction of other pathways compared with latter time points. Interestingly, treatment with the unarmed virus (VV CTRL) had a limited impact on the tumor transcriptome in CT26 model but a stronger effect in B16F10 model, with about half of treated mice showing a strong activation of immune pathways compared with none of the untreated mice and all mTG6050 treated mice (figure 5A). To assess the effect of IL-12 expression, markers of the type II IFN pathway31 were analyzed in more detail. Most of them were significantly upregulated after treatment with mTG6050 in the CT26 model (online supplemental figure S23). In the B16F10 model, most markers were upregulated treatment with either VV CTRL or mTG6050, however, no specific differentiating transcriptomic signature was observed between these two viruses. Interestingly, the IL-12 receptor subunits Il12rb1 and Il12rb2 were overexpressed in tumors treated with mTG6050 compared with VV CTRL in both models (online supplemental figure S24).

Figure 5

Transcriptional analysis shows significant changes in the transcriptome of tumors after treatment with mTG6050. (A) Heatmaps showing gene set variation analysis enrichment scores for markers of immune components and pathways in tumors (n=10) treated with buffer, 1E+07 PFU/injection VV-Luc (VV CTRL) or mTG6050 in two tumor models: CT26 (D11, D15 and D17; left panel) and B16F10 (D11 and D14; right panel). Data were generated following 3’ RNA sequencing. (B) UMAP showing clusters of cell populations assigned using cell type markers, from CT26 tumors treated with buffer (left) or 1E+07 PFU/injection mTG6050 (right) at D11. (C) UMAP generated after in silico removal of tumorous cells and CAFs and re-clustering of remaining immune cells. Cell types inferred using markers are indicated by labels. Left panel: cells expressing Cd8a and Foxp3 are colored in red and blue, respectively. Middle panel: cells showing high expression of a set of three macrophages M1 markers (Stat1, Socs1 and Nos2) are colored in red. Right panel: cells showing high expression of a set of three macrophages M2 markers (Stat6, Socs2 and Arg1) are colored in blue. Cells expressing the gene sets M1 and M2 were identified using the R package AUCell. Percentages of cells expressing Cd8a, Foxp3, M1 or M2 markers are reported in the table below, along with the ratio Cd8a/Foxp3 and M1/M2. APC, antigen-presenting cells; CAF, cancer-associated fibroblasts; IFN, interferon; MHC, major histocompatibility complex; NK, natural killer; NKT, natural killer T cells; pDC, plasmacytoid dendritic cells; PFU, plaque forming unit; UMAP, Uniform Manifold Approximation and Projection.

Cd8a expression, a marker of effector T cell, increased significantly after treatment with mTG6050 at all time points in both models (online supplemental figure S25). In the B16F10 model, Foxp3 expression, a marker of Treg, largely increased in VV CTRL-treated mice but was kept stable with mTG6050 treatment. In the CT26 model, Foxp3 expression remained stable after treatments (online supplemental figure S25). This data suggested a shift towards a higher CD8+T cells over Treg in the tumors. Interestingly, PD-L1 (Cd274) was also more expressed in tumors treated with mTG6050, compared with VV CTRL or vehicle, in both models (online supplemental figure S25).

ScRNA sequencing confirmed the tumor immune infiltration (figure 5B and online supplemental figure S26). To profile immune cells in more detail, tumor cells and cancer-associated fibroblasts were in silico removed, and the remaining cells were re-clustered (figure 5C and online supplemental figure S27). Analysis of Cd8a and Foxp3 markers showed a strong increase in the CD8+/Treg ratio, (figure 5C). Additionally, the gene signature of M1 and M2 macrophage markers (three markers each) also revealed an increase of M1 over M2, confirming the 3’ RNA sequencing data, and flow cytometry results that innate and adaptive immune pathways are impacted by mTG6050 treatment (figure 5C). Moreover, IL-12R and PD-L1 transcripts were increased in the TME of mice treated with mTG6050 compared to mice in the control groups (online supplemental figure S28). The increase in IL-12R and PD-L1 might correspond to a positive feedback loop on expression of IL-12 and IFN-γ proteins, respectively.

Combination with ICI improved the antitumoral activity of mTG6050

Following these results, we hypothesized that combining mTG6050 with an anti-murine PD-1 (@mPD-1) could potentiate the antitumor response induced by mTG6050 (figure 6A). Although the combination of mTG6050 with @mPD-1 did not significantly improve survival compared with mTG6050 alone in any of the three models, a trend toward higher efficacy for the combination was observed: that is, 6/10 survival rate versus 2/10 in CT26, 6/10 versus 3/10 in B16F10, and 2/10 versus 0/10 in LLC1 (figure 6B). Considering these results and the mode of action (MOA) of TG6050, the combination with an immune checkpoint inhibitor (@PD-1) was envisaged as an option for the upcoming clinical development.