A novel oncolytic virus engineered with PD-L1 scFv effectively inhibits tumor growth in a mouse model

The adoptive transfer of engineered chimeric antigen receptor (CAR) T cells has yielded impressive clinical results for targeting hematological cancers.¹ However, the advancement of CAR T-cell therapy in solid tumors remains limited, due to the scarcity of tumor antigens that are deemed safe for targeting. Moreover, the requirement of patient-specific autologous T cells for the current standard treatment has also hampered its application. Oncolytic virus (OV) therapy has been classified as another form of novel immunotherapy. In particular, Talimogene laherparepvec (T-VEC, Imlygic), a genetically modified herpes simplex virus expressing GM-CSF, is the first OV-based drug approved by the US Food and Drug Administration.² In addition, monoclonal antibodies targeting immune checkpoint molecules have also recently been used to make substantial progress in the treatment of cancer therapy. In particular, antibodies blocking the programmed death-1 (PD-1)/programmed death ligand-1 (PD-L1) pathway have displayed promising results in a wide variety of cancers and other diseases.^3–5 A recent study found that T-VEC virotherapy combined with anti-PD-1 resulted in a high response rate in patients with metastatic melanoma in a phase 1b trial.⁶ Thus, in this study, we sought to generate an OV bearing an antibody against an immune checkpoint, PD-L1, and evaluated its ability to modulate anticancer effects in mice as well as in the immunosuppressive tumor microenvironment.

Vesicular stomatitis virus (VSV) preferentially replicates in cancer cells, which commonly exhibit defects in type I IFN signaling, making it an ideal oncolytic immunotherapy vector. Although laboratory strains of recombinant VSV are not usually pathogenic in humans, an attenuated VSV was generated as a therapeutic vector to avoid the potential for VSV to cause disease in humans. An M51R substitution in the VSV M gene has been reported to prevent VSV from inhibiting host gene expression, thus allowing cells to mount an antiviral response, as previously described.⁷ Therefore, we generated an M51R mutation in the M gene of the VSV viral backbone, and the attenuated virus (VSV^M51R) was recovered in BHK cells.⁸ The viral replication curve indicated that VSV^M51R replicated more slowly than the WT VSV-GFP virus in 293T cells, indicating its attenuation (Supplementary Data S1A). In Vero cells, which are type I IFN-deficient, VSV^M51R replication was comparable with that of VSV-GFP (Supplementary Data S1B). Moreover, in mice intranasally infected with 10⁷ pfu of the virus, VSV^M51R exhibited substantially attenuated toxicity in vivo (Supplementary Data S1C). Taken together, these data indicate that we successfully generated an attenuated VSV^M51R immunotherapy vector.

A synergistic effect has been observed for combination therapy with T-VEC and immune checkpoint antibodies.⁶ However, little is known about the oncolytic effects of a recombinant OV expressing a PD-1/PD-L1 antibody compared with that of combination therapy. Therefore, we engineered a VSV^M51R expressing a single-chain antibody Fv fragment (scFv) encoded by the PD-L1-targeting antibody, avelumab. Avelumab is a human IgG1 antibody with antibody-dependent cell-mediated cytotoxic activity developed by Merck (Darmstadt, Germany) and Pfizer that is now in multiple phase III clinical trials for the treatment of non-small-cell lung cancer (NCT02395172), advanced renal cell cancer (NCT02684006), and gastric cancer (NCT02625610). We obtained the coding sequence of avelumab scFv, as previously described.⁹ There is a (GGGS)₄ linker between the light chain, heavy chain, and His tag in the C terminus of the scFv. The synthesized scFv fragment was inserted between the VSV G and L proteins using XhoI and NheI restriction enzymes (Supplementary Data 2A). The recombinant virus expressing avelumab scFv (VSV^M51R-PD-L1) was recovered in BHK cells, and scFv expression was further confirmed by a western blot of the His tag fused to the scFv in VSV^M51R-PD-L1-infected Lewis lung carcinoma (LLC) and B16 tumor cells (Supplementary Data 2B).

The amino acid identity between human PD-L1 (hPD-L1) and mouse PD-L1 (mPD-L1) is 69%, as calculated by BLAST. To generate a mouse tumor model, mouse PD-L1 was first knocked out in B16 cells using CRISPR/Cas9 (Supplementary Data 3A). Next, we expressed hPD-L1 in mPD-L1 knockout cells, and the cells expressing hPD-L1 were sorted via flow cytometry (Supplementary Data 3B). Finally, we incorporated the luciferase gene into hPD-L1 knock-in B16 cells by lentiviral transduction. Similarly, we also obtained hPD-L1 knock-in LLC cells with the mCherry reporter gene. Using a competition ELISA, the relative concentration of scFv in infected LLC cell culture medium was found to be ∼50 ng/mL (Supplementary Data 3C). It has been reported that treatment with the HDAC inhibitor panobinostat in multiple myeloma immunotherapy upregulates PD-L1 expression.¹⁰ We found that PD-L1 expression was upregulated upon VSV^M51R-PD-L1 infection in vitro and in vivo (Supplementary Data 4A, B).

Mice were subcutaneously injected with 4 × 10⁵ hPD-L1 LLC cells to generate a tumor model. The mice intratumorally received 10⁷ pfu VSV^M51R-PD-L1 8 days post inoculation, three times every 2 days (Supplementary Data 5A). The expression of hPD-L1 scFv in the injected mice greatly reduced the tumor size. The average tumor size in the VSV^M51R-PD-L1 group was 170 mm³, while it was 446 mm³ in the VSV^M51R group 18 days post inoculation (n = 15) with LLC. Similar trends were observed using murine melanoma B16 cells (Supplementary Data 5B). Accordingly, compared with VSV^M51R treatment, VSV^M51R-PD-L1 treatment greatly improved the mouse survival rate from 20 to 60% after LLC inoculation (Supplementary Data 5C). Although all mice in the B16 inoculation groups died, we still observed that VSV^M51R with PD-L1 scFv could prolong the survival of mice. To our surprise, in immune-deficient Balb/c nude mice, there were no differences between PBS-treated and either VSV^M51R- or VSV^M51R-PD-L1-treated mice. The average tumor size was 1385, 1049, and 1138 mm³ in the PBS group, VSV^M51R group, and VSV^M51R-PD-L1 group, respectively (Supplementary Data 6A, B). This finding indicates that the host antitumor immune response plays a notable role in the control of the tumor.

We then compared the therapeutic effects of VSV^M51R-PD-L1 with the combination therapy of hPD-L1 avelumab scFv + VSV^M51R (Fig. 1a) and found that the therapeutic effect of VSV^M51R-PD-L1 was comparable with that of PD-L1 scFV + VSV^M51R combination therapy. It is important to note that this effect was still significantly superior to that of either PD-L1 scFv or VSV^M51R (Fig. 1b). Interestingly, we found that the growth of tumors was obviously inhibited in the combination therapy (6, n = 9) and the VSV^M51R-PD-L1 groups (5; n = 9), which is consistent with the objective rate (62%) found in a clinical trial using T-VEC and pembrolizumab combination therapy.⁶ Accordingly, the mice that received the VSV^M51R-PD-L1 or PD-L1 scFv + VSV^M51R combination therapy displayed the highest survival rate following LLC cell inoculation (Fig. 1c). To investigate the antitumor memory immune response induced by VSV^M51R-PD-L1 therapy, we rechallenged the cured mice with LLC tumor cells. Normal C57BL/6 WT mice were used as a control group. A tumor formed in the groin of only one of the six cured mice, and only increased to 136 mm³ 30 days post inoculation. In contrast, tumors formed in all of the mice in the control group, and these mice died 36 days post inoculation (Fig. 1d). The total number of CD8 central memory (CD44⁺CD62L⁺) T cells and CD8/CD4 effector memory (CD44⁺CD62L⁻) T cells was dramatically increased in the spleens of the VSV^M51R-PD-L1 cured mice compared with that in the spleens of the normal mice (Supplementary Data 7). In addition, we found that the IFN-γ⁺ CD8 T cells, which represent the magnitude of the CD8 T-cell expansion,¹¹ were also significantly increased in the spleens of the cured mice, suggesting that VSV^M51R-PD-L1 induced an antitumor memory immune response that protected the mice from recurrence (Fig. 1e).

Fig. 1.

The therapeutic effect of VSV^M51R-PD-L1 compared with that of VSV^M51R and PD-L1 scFv combination therapy for lung cancer. a A schematic depicting the OV treatment schedule. b Distribution of tumor volume on day 18 after implantation. Error bars represent the standard error of the mean. c The survival analysis tested by log-rank (Mantel–Cox) test for different treatments. d Mice (n = 6) cured of LLC^hPD-L1 tumors with VSV^M51R-PD-L1 were rechallenged with LLC cells. The graph shows the percentage of tumors detected before the animals reached the endpoint. e Mice bearing LLC^hPD-L1 tumors that were treated with VSV^M51R-PD-L1 and remained tumor-free for 200 days were rechallenged with LLC^hPD-L1 tumor cells (2 × 10⁵) implanted subcutaneously in the right flank. Normal mice implanted with 2 × 10⁵ LLC^hPD-L1 tumor cells were used as controls (n = 4). f Spleens were isolated from VSV^M-PD-L1 survivor and control mice 9 days post LLC cell inoculation. T cells from the spleen were collected, stained with anti-CD8, and stained for intracellular expression of IFN-γ. *p < 0.05

In conclusion, our study shows that VSV^M51R engineered with PD-L1 scFv exhibits potent therapeutic effects, similar to combination therapy with a PD-L1 avelumab scFv in mouse tumor models. Although the safety and therapeutic efficacy of VSV^M51R-PD-L1 require further testing in clinical studies, our results suggest that VSV^M51R-PD-L1 is a promising reagent for cancer therapy.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version of this article (10.1038/s41423-019-0264-7) contains supplementary material.