Modulation of the tumor microenvironment by armed vesicular stomatitis virus in a syngeneic pancreatic cancer model

Pancreatic Ductal Adenocarcinoma (PDAC) represents ~ 90% of all pancreatic cancers [1]. While surgery can be curative in some PDAC patients when disease is diagnosed at early stage, most PDAC patients present with metastasis or are not operable at the time of diagnosis due to locally advanced disease [2, 3]. Several front-line chemotherapies used as the standard of care for advanced and metastatic PDAC resulting in overall survival benefits measuring only in months [4, 5]. Key mutations in oncogenes and tumor suppressor genes have been identified in PDAC including Kirsten-ras protein (KRAS), cyclin-dependent kinase inhibitor 2A (CDKN2A), tumor protein p53 (TP53) and mothers against decapentaplegic homolog 4 (SMAD4) [6]. However, targeting these pathways is plagued by resistance and limited availability of useful targeted agents. Given these factors, improved treatment strategies for PDAC are urgently needed. Gemcitabine is the most commonly used drug since 1997. More recently, combination therapy of 5-fluoruracil (5-FU), leucovorin (LV), irinotecan and oxaliplatin (FOLFIRINOX) and gemcitabine-nab-paclitaxel significantly increased the median overall survival in comparison with gemcitabine [7]. A large number trials with different combination therapies, with addition of drugs targeting new pathways, have not yet demonstrated further improvement [8]. These efforts are ongoing with addition of more targeted therapies. Due to the complicated genetic plasticity of PDAC, it is very hard to design an effective combination therapy. While improvements along these strategies are likely to continue, several new approaches have emerged lately.

To date, immunotherapy has also shown limited success in PDAC patients [9]. This limited efficacy is a consequence of multiple factors. First, an intense stromal desmoplastic reaction can limit access of cells into tumors. This occurs possibly both by physical barriers due to excess collagen production and via secretion of soluble factors, including CXCL12, that prevent proper T cell chemotaxis [10]. Second, many cells within the tumor microenvironment (TME) serve as a source of immunosuppressive cytokines including interleukin-6 (IL-6) and transforming growth factor-β (TGF-β) that further inhibit antitumor immune responses via redundant mechanisms [11‐13]. Third, abundant suppressive immune cells including T regulatory cells (T reg), myeloid derived suppressor cells (MDSC) are also present that further maintain “immune privilege” [14‐16]. Finally, PDAC tumors have a lower frequency of somatic mutations, which in theory limits the abundance of neoantigens available for promoting immune response [17]. To circumvent these multiple limitations, combination therapies consisting of chemotherapy or radiation with immunotherapy are currently under investigation in several clinical trials for patients with PDAC [18].

A newly emerging therapy using oncolytic virus (OV) has shown improvements on the efficacy of immunotherapy in pre-clinical models [19, 20]. OV can change the TME to enhance anti-tumor immunity and releases novel neoantigens through its oncolysis. It was shown that tumor-associated macrophages (TAMs), especially the anti-inflammatory macrophages, were downregulated and increased the percentage of tumor-infiltrating lymphocytes. Activated cytotoxic CD8 + T cells and T helper (Th)1 cells were increased by treatment of OV herpes simplex virus-1 in the syngeneic PDAC model, based on single cell RNA sequencing (scRNA-seq) and multicolor fluorescence-activated cell sorting (FACS) analysis [19]. A significant enhancement of tumor-specific IFNγ production was observed by restimulation with growth-arrested tumor cells of splenocytes isolated from murine pancreatic DT6606 subcutaneous tumors treated with a novel oncolytic Vaccinia virus [21]. As more evidence for this, the efficacy of anti-PD-1 immunotherapy in treatment of patients with advanced melanoma was enhanced when combined with talimogene laherparepvec (T-Vec) [22]. Patients who responded to combination therapy had increased CD8 + T cells, and elevated PD-L1 protein expression in tumors. We have devised an armed oncolytic virus, VSV-S, that has an expression cassette for Smac inserted in the genome of vesicular stomatitis virus (VSV). Smac is a mitochondrial protein that mitigates a class of negative regulators of apoptosis, known as the inhibitor of apoptosis proteins (IAPs), when Smac is released in the cytosol. The endogenous Smac was diminished by infection of wt VSV [23]. VSV-S expressed a high level of Smac to replenish the endogenous Smac and induced elevated apoptosis via the caspase-9 pathway and strong tumor necrosis in a human breast cancer model in nude mice, and inhibition of tumor growth in the syngeneic mouse model [23]. We report here the change of the TME induced by VSV-S in the KPC-based mouse model of pancreatic cancer, and demonstrate the combined treatment with VSV-S and anti-PD-1 antibody has significant growth inhibition as compared to either agent alone.

Methods that turn PDAC from immunologically “cold” to immunologically “hot” are likely to improve the outcome of immunotherapy such as immune checkpoint blockade. Among numerous approaches, treatment with OV may have some unique advantages. Infection of tumor cells by OV can stimulate inflammatory responses in the TME and release quantities of tumor antigens to elicit anti-tumor immunity. A number of OVs have been evaluated in Phase II trials, including adenovirus (ONYX-015) [28], reovirus (Reolysin) [29] and parvovirus (ParvOryx) [30]. While treatment with OVs, or in combination with gemcitabine, was well tolerated, the positive outcomes are encouraging in some cases, but not clearly conclusive in the overall study [31].

To design a robust OV that potently targets cancer cells, we constructed VSV-S derived from a rapidly replicating virus, vesicular stomatitis virus. VSV expressing INFβ (VSV-IFNβ-NIS) has been investigated in a phase I-II study in patients with refractory solid tumors, showing a good safety profile, and is currently evaluated in combination with cemiplimab [32]. In our study, VSV-S expresses the Smac protein, a pro-apoptotic protein, during its infection. Our previously published data have shown that VSV-S maintains a high level of intracellular Smac and induced a high degree of apoptosis in VSV-S infected cells, whereas infection by the wt VSV diminished endogenous Smac and has a more prominent resistance by several cancer cell lines [23]. In addition to the robustness of VSV-S in killing cancer cells, we developed here a strategy to enhance cancer targeting of VSV-S through adaptation by limited dilution. VSV has a broad cell tropism [25], which permits sufficient initial infectivity in the selected target cell. Since VSV replicates rapidly in most cells, targeting the selected cell can be achieved through adaptation, an efficient approach commonly used for increasing virus selectivity [33]. Other groups have adapted oncolytic VSV to improve replication efficiency (i.e. virus yield) [34]. Our adaptation strategy is using limited dilution to increase selective infection, rather than a high virus yield. In each round of infection, the inoculum virus was selected from the supernatant of cells infected with the highest dilution in the previous round. By a few rounds of adaptation under such a strategy, the selective infectivity of VSV-S was increased by 100-fold in MIA PaCa-2 cells (Table 1), a human PC cell line, and 20-fold in mouse KPC cells (Table 2). To directly confirm the selective infectivity, infection of KPC and MS1 cells by adapted VSV-S was compared by their CC₅₀ values measured with MTT assays (Fig. 1). The selectivity is 94 fold for KPC cells over MS1 cells. Our results demonstrated that VSV-S adaptation by limited dilution is readily applicable to target other cancer cells in a very short period of time (a few days).

The ultimate application of VSV-S or its derivatives is to enhance the efficacy of treatment for PDAC and other cancers. Previously, we have shown that VSV-S has a superior antitumor activity in animal models in comparison with wt VSV [23]. Published animal studies also confirmed that the efficacy of anti-PD-1 therapy alone is very limited [35‐38]. Our primary interest is to optimize the combination therapy of OV and checkpoint inhibitors, especially when adapted VSV-S changed the TME significantly. The study using the syngeneic mouse model for PDAC was carried out to investigate adapted VSV-S. Overall, intratumoral injection of VSV-S_PKC steadily inhibited tumor growth, occasionally tumors were eliminated (2 out of 8 female mice). Treatment with anti-PD-1 antibody, 8 days after virus injection, prolonged inhibition of tumor growth or even helped to eliminate the tumor (3 out of 8 female mice, and 2 out of 8 male mice). Although we did not include an anti-PD-1 antibody-only group as in previously published reports, treatment by anti-PD-1 antibody alone did not change tumor growth or survival in KPC mouse model as reported in [39] and survival was increased by combination therapies [36]. The delayed administration of the immune checkpoint inhibitor in combination with OV is consistent with the temporal change of the TME, and is cooperated with results from similar studies [40]. The combination therapy greatly extended the survival of tumor bearing mice, achieving a long-term survival of 44% (7 out of 16 subjects).

The efficacy of the treatment with OV appeared to be dependent on the initial tumor size when OV treatment began. This could be due to the degree of virus spread within the tumor mass. When the tumor size is smaller, it is expected that a higher portion of the tumor would be directly exposed to OV infection upon intratumoral injection. It is also observed that the growth rate of tumors in female mice after treatment was significantly slower than that in male mice. The responses to VSV-S/anti-PD-1 treatment are also more favorable in female mice versus male mice. It is not clear what the observation really means in this limited study. The patient numbers of incident cases and deaths peaked at the ages of 65–69 years for males and at 75–79 years for females [41]. PDAC appears to be more devastating to males than females. It might be possible that our preliminary data could suggest that female patients may respond better to virotherapy.

To achieve tumor regression, the rate of cancer killing by OV and TIL must overtake that of tumor growth. In this subcutaneous model, treatment with OV modestly reduced the rate of tumor growth. More importantly, however, treatment with OV altered the TME that could enhance the efficacy of immunotherapy. This advantage is illustrated by large changes of leucocytes in the tumors infected by VSV-S (Fig. 2). The change was noticeable after 2 days, and more pronounced after 8 days, following the final VSV-S infection. Despite that the total number of lymphocytes was greatly increased, the ratio between lymphocytes and myeloid cells, or between CD8 + and CD4 + T cells, however, was not changed significantly. The most prominent changes of myeloid cell compartments were the dramatic increase of neutrophils, and the decrease of MDSCs and macrophages. These changes are consistent with strong inflammation in the tumor caused by VSV-S infection. Neutrophils exhibit tumor suppressive activities by generating reactive oxygen species (ROS), activation of the IFN-γ pathway, and up-regulation of antigen presentation [42]. The large increase of neutrophils in tumors could cause more death of cancer cells due to innate immunity [43]. Several studies confirmed that neutrophils participated in tumor cell clearance upon OV infection of tumors [44‐47], which is also consistent with our previous results of efficacious tumor regression by VSV-S treatment of breast cancer xenografts [23].

The large reduction of macrophages, especially M2-like macrophages (Fig. 2e, f), most likely were related to reverting the immunosuppressive TME. To confirm this notion, levels of ARG1, TGF-β, and IL-10 in tumors were analyzed by immunohistochemistry (Fig. 2g). The association of their expression levels with the immune conditions in the TME is not always clear cut. For instance, overexpression of arginase is perceived as a poor prognostic factor in a wide variety of cancer types [48]. Myeloid cells are major contributors to immune defense against pathogens and play an important role in tissue remodeling. In the tumor, myeloid cells are highly heterogeneous. Cells associated with strong immunosuppressive functions are mainly MDSCs that express arginase. It has been shown that arginase plays an opposite role in the immune response and is one of the main mechanisms of immunosuppression [49]. MDSCs also express suppressive cytokines like TGF-β and IL-10 in the tumor [49]. Our observation is consistent with changing the immunosuppressive TME by VSV-S infection of the tumor. The observed reduction in PD-L1 may be associated with its reduction in myeloid cells, not necessarily PD-L1 expressed in tumor cells. More extended regression of tumors should be achievable by optimal combination regimens of OV treatment and immunotherapy.

Preliminary data of clinical chemistry, hematology and coagulation were collected in tumor bearing and control mice after 1.0 × 10⁸ PFU of VSV-S_KPC was intravenously injected (Table 3). There is no indication of safety concerns. Our data are similar as those obtained for another modified VSV [50]. Treatment with VSV-S increased white blood cells (WBC) in mice without tumors. Lymphocyte counts in tumor bearing mice were reduced, whereas they are normal in control mice. Neutrophils and monocytes were largely increased in tumor bearing mice than mice with tumors. The changes in these cells are due to tumors in mice, not treatment of VSV-S. No other significant side effects were observed.

Our results demonstrated that infection by VSV-S dramatically changed the immune cell distribution in the TME. The TME changes are consistent with the beneficial outcome of combination therapy with anti-PD-1 antibody. More detailed immunological studies may be extended from the current data to further optimize the combination therapy in orthotopic PDAC models.