Rational combination of oncolytic vaccinia virus and PD-L1 blockade works synergistically to enhance therapeutic efficacy

Abstract

Both anti-PD1/PD-L1 therapy and oncolytic virotherapy have demonstrated promise, yet have exhibited efficacy in only a small fraction of cancer patients. Here we hypothesized that an oncolytic poxvirus would attract T cells into the tumour, and induce PD-L1 expression in cancer and immune cells, leading to more susceptible targets for anti-PD-L1 immunotherapy. Our results demonstrate in colon and ovarian cancer models that an oncolytic vaccinia virus attracts effector T cells and induces PD-L1 expression on both cancer and immune cells in the tumour. The dual therapy reduces PD-L1⁺ cells and facilitates non-redundant tumour infiltration of effector CD8⁺, CD4⁺ T cells, with increased IFN-γ, ICOS, granzyme B and perforin expression. Furthermore, the treatment reduces the virus-induced PD-L1⁺ DC, MDSC, TAM and Treg, as well as co-inhibitory molecules-double-positive, severely exhausted PD-1⁺CD8⁺ T cells, leading to reduced tumour burden and improved survival. This combinatorial therapy may be applicable to a much wider population of cancer patients.

Figure 1. PD-L1 is elevated post vvDD infection in vitro.

(a) MC38-luc colon cancer or ID8-luc ovarian cancer cells (4 × 10⁵ cells each) were mock-infected or infected with vvDD. These cells were harvested 24 h later, blocked with α-CD16/32 Ab and then stained with α-PD-L1 for flow cytometry. (b) Total RNA was extracted from the harvested tumour cells and used in RT–qPCR to determine PD-L1 expression. In a, ISO: isotype IgG control used for staining; CM: condition medium. Data are presented as individuals, mean +/− s.d. (c) Cells from a panel of human cancer cell lines representing colorectal, ovarian, lung, cervical cancer and mesothelioma were infected or mock-infected with vvDD for 24 h and total RNA was prepared and subjected to RT–qPCR to determine the relative expression of PD-L1. The values on the y axis indicate the ratio of PD-L1 expression between infected versus mock-infected cancer cells. Data are presented as mean +/− s.d.

Figure 2. PD-L1 is elevated in cancer tissue post vvDD treatment in vivo.

B6 mice were subcutaneously inoculated with MC38-luc cells (4 × 10⁵ per mouse). PBS or vvDD was intratumorally injected at 2 × 10⁸ pfu per tumour when the s.c. tumour area reached 5 × 5 mm². Tumour tissues were collected from PBS or virus-treated mice 4 days post treatment. Collected tumour tissues were weighed and incubated in RPMI 1640 medium containing 2% FBS, 1 mg ml⁻¹ collagenase, 0.1 mg hyaluronidase, and 200 U DNase I at 37 °C for 1–2 h to make single cell preparations. The single cells were used for extraction of total RNA for RT–qPCR assay (a), or they were blocked with α-CD16/32 Ab and then stained with antibodies against CD45, CD11b, Gr1, PD-L1, F4/80 to determine PD-L1 expression on tumour cell (CD45⁻) (b), MDSC (CD45⁺CD11b⁺Gr1⁺) (c) and TAM (CD45⁺CD11b⁺F4/80⁺) (d). In this experiment, the anti-PD-L1 Ab clone 10F.9G2 was used for analysis. Significant differences are indicated by *(P<0.05) or ** (P<0.01) determined by t-test. In this and other figures, the standard symbols for P values are *P<0.05; **P<0.01.

Figure 3. Primary tumour growth is dynamically monitored post treatments.

B6 mice were intraperitoneally inoculated with 5 × 10⁵ MC38-luc cancer cells and treated as described. Tumour-bearing mice were killed on days 2, 5 and 13 after first treatment, and primary tumour tissues were collected. Shown are schematic representations of the experiment (a) or weight of primary tumours on day 2 (b), day 5 (c) and day 13 (d) post first treatment. Numbers of mice per group are 5–11 (n=5∼11). In photographs, only representative three tumours from each group are shown. In the figures, VV=vvDD-CXCL11. Data were presented as individuals and means. Student's t-test was used to analyse the statistical significance (*P<0.05; **P<0.01; ***P<0.001; ****P<0.0001; and NS: not significant).

Figure 4. VV and PD-L1 blockade synergistically elicit anti-tumour effect.

B6 mice were intraperitoneally inoculated with 5 × 10⁵ MC38-luc or 3.5 × 10⁶ ID8-luc cells, respectively, and divided into required groups according to tumour growth condition based on live animal IVIS imaging 5 days post tumour cell injection. Grouped mice were intraperitoneally injected with VV (2 × 10⁸ pfu per 100 μl), α-PD-L1 Ab, VV plus α-PD-L1 Ab, or PBS (100 μl) per mouse, respectively. α-PD-L1 Ab (200 μg per 100 μl) was injected every 2 days for a total of four times, according to the scheme in Fig. 3a. Live animal imaging of the mice with MC38-luc tumours at day 9 post first treatment (a). The data are quantified (n=8∼10) (b). The survival of tumour-bearing mice was monitored by Kaplan–Meier analysis and statistical analyses were performed with Log rank test. Both MC38-luc (c) and ID8-luc (d) -tumour-bearing mice were shown.

Figure 5. The α-PD-L1 treatment reduces the PD-L1⁺ cells in TME.

B6 mice were intraperitoneally inoculated with 5 × 10⁵ MC38-luc cancer cells and treated with VV and/or α-PD-L1 as described. Tumour-bearing mice were killed at day 5 post first treatment and primary tumours were collected and analysed to determine the PD-L1⁺ CD45⁻ cells (a,b), PD-L1⁺ DC (defined as CD45⁺ CD11b⁺CD11c⁺Ly6G⁻PD-L1⁺) (c), PD-L1⁺G-MDSC (defined as CD45⁺CD11c⁻CD11b⁺Ly6G⁺Ly6c^lowPD-L1⁺) (d), PD-L1⁺ M-MDSC (defined as CD45⁺CD11c⁻CD11b⁺Ly6G⁻Ly6c^hiPD-L1⁺) (e), PD-L1⁺ TAM1 (defined as CD45⁺CD11c⁻CD11b⁺Ly6G⁻F4/80⁺CD206⁻PD-L1⁺) (f), PD-L1⁺ TAM2 (defined as CD45⁺CD11c⁻CD11b⁺Ly6G⁻F4/80⁺CD206⁺PD-L1⁺) (g). Of note, the anti-PD-L1 antibody clone 10F.9G2 was used for therapy while clone MHI5 was used for subsequent analysis. Data were analysed using Student's t-test (*P<0.05; **P<0.01; ***P<0.001; ****P<0.0001).

Figure 6. The combination of VV plus α-PD-L1 treatment enhances infiltration of effector T cells and reduces Treg cells and exhausted CD8⁺ T cells in the TME.

B6 mice were intraperitoneally inoculated with 5 × 10⁵ MC38-luc and treated with VV and/or α-PD-L1 as described. Single cells were made from primary tumours collected from tumour-bearing mice at day 5 post first treatment, blocked with α-CD16/32 Ab and then stained with antibodies against CD45, CD8, CD4, PD-1, ICOS, PD-1, CTLA-4, TIM-3, LAG-3, TIGIT and Foxp3 to determine the quantities of CD8⁺ T cells (a), CD8⁺ T-cell activation (b–d), CD8⁺ T-cell exhaustion (e–h), Treg cells (i), CD8⁺/CD4⁺Foxp3⁺ T cells (j) and CD4⁺Foxp3⁻ T cells (k,l) in the TME. Of note, the anti-PD-L1 antibody clone 10F.9G2 was used for therapy while clone MHI5 was used for subsequent phenotypic analysis. Data were analysed using Student's t-test (*P<0.05; **P<0.01; ***P<0.001).

Figure 7. The combination therapy leads to enhanced activation markers in the TME.

In one experiment, the i.p. MC38-luc tumour model was treated with VV and/or α-PD-L1 as described. Tumour tissues were collected at day 5 post first treatment and were used for extraction of total RNA. RT–qPCR assays were performed to determine the levels of IFN-γ (a), granzyme B (b) and perforin (c) in the TME. Data were analysed using Student's t-test (*P<0.05; **P<0.01).

Figure 8. The systemic anti-tumour immunity elicited by dual therapy plays an important role in the overall therapeutic efficacy.

(a) B6 mice were intraperitoneally inoculated with 5 × 10⁵ MC38-luc cancer cells and treated with VV and/or α-PD-L1 as described. Splenic CD8⁺ T cells (4 × 10⁵) were isolated from naive and MC38-luc-bearing mice that received different treatments 18 days post tumour cell injection and restimulated with mitomycin C-treated MC38-luc or B16 cancer cells (4 × 10⁴ cells each) in the presence of 4000-rad-irradiated CD8-depleted naive B6 splenocytes (2 × 10⁶) in 200 μl RPMI-1640 medium supplemented with 10% FBS at 37 °C, 5% CO₂ for 2 days. The concentration of IFN-γ in the culture supernatants was tested by ELISA. The statistical analyses were performed with t-test. (b) Naive or MC38-luc-bearing B6 mice with dual treatments, which survived for more than 60 days, were s.c. rechallenged with 1 × 10⁶ MC38-luc cancer cells. The primary tumour size was measured and presented here. (c) In a separate experiment, B6 mice were inoculated with 5 × 10⁵ MC38-luc cells i.p. and treated with VV plus α-PD-L1 or PBS at day 5 post tumour inoculation, α-PD-L1 Ab was injected every 2 days for a total of four times. α-CD8 Ab (250 μg per injection), α-CD4 Ab (150 μg per injection) or α-IFN-γ Ab (200 μg per injection) were intraperitoneally injected into mice to deplete CD8+ T cells, CD4+ T cells or neutralize circulating IFN-γ as scheduled in c, and the overall survival was monitored by Kaplan–Meier analysis and analysed using log rank test (d).