Integrating innate and adaptive immunity in oncolytic virus therapy

Abstract

Oncolytic viruses, viruses engineered to lyse tumor cells, work hand in hand with the immune response. While for decades the field isolated lytic capability and viral spread to increase response to virotherapy, there is now a wealth of research that demonstrates the importance of immunity in the oncolytic virus mechanism of action. In this review we will cover how oncolytic viruses interact with the innate immune system to fully activate the adaptive immune system and yield exceptional tumor clearances as well as look forward at combination therapies which can improve clinical responses.

INTRODUCTION

The cancer immunity cycle is now a well understood part of cancer progression and therapy¹. At each stage of this cycle, the tumor has developed resistance mechanisms which inhibit the ability of the immune system to clear them. These include downregulation of tumor antigen and MHC, physical barriers which reduce T cell infiltration, inhibitory receptors which reduce T cell activity, and a myriad of suppressive cytokines, cell types, and metabolites which reduce anti-tumor immunity. Fortunately, we have developed therapies which target each of these resistance mechanisms, to varying degrees of success. Many techniques can be utilized for successful regression; stimulation of an immune response if none is present, enhancing the existing response through targeting suppressive mechanisms, redirecting the immune response to tumor antigens, or enhancing antigen expression on the tumor if it has been lost. Excitingly, all of these are achieved with successful oncolytic virus therapy.

Oncolytic viruses (OVs) are DNA or RNA viruses that have been modified or selected for their tumor specificity. OVs are injected either intravenously or directly into the tumor. They then selectively enter the tumor cell, through engineered selectivity mechanisms, where they replicate and lyse the tumor, leading to immunogenic cell death (ICD, see Glossary), viral spread, as well as release of pathogen associated molecular patterns (PAMPs) and damage associated molecular patterns (DAMPs), and other immune stimulating factors. The virus may also trigger innate viral sensing in the tumors upon entry and replication. Together, this leads to the stimulation of a strong adaptive immune response to both viral and tumor antigen and can lead to complete tumor eradication, tumor shrinkage at distal uninjected sites, and ultimately immune memory. OVs can also be engineered to deliver “payload”: inserting a gene of interest into the virus results in production of the gene as the virus replicates. This can further enhance the therapeutic response to OVs as multiple mechanisms can be targeted in the same therapeutic dose.

While OV research initially focused heavily on the replicative and lytic capabilities of the virus and suggested that viral clearance by the adaptive immune system would hinder OV therapy, there is now a breadth of data showing the importance of the immune system in OV-mediated tumor clearing. More recently, the importance of avoiding triggering innate sensing mechanisms and viral lysis have been called into question. Here, we review the most recent data regarding OV mechanism of action, from tumor cell entry to CD8+ cytotoxic T cell killing, and succinctly outline the importance of the immune system at each step (Fig. 1).

Figure 1: Oncolytic viruses and the immune system.

1. The oncolytic virus enters the tumor cell and begins replication. Viral RNA or DNA is recognized by sensors, cGAS or RIG-I/MDA5, and this leads to downstream signaling and activation of transcription of ISGs, type I interferons, and cytokines. Viral replication continues to assembly and finally, the tumor cell is lysed and viral particles are released and spread throughout the tumor. 2. Innate sensing also increases immunogenic cell death (ICD) which leads to the release of PAMPs and DAMPs like HMGB1, ATP, and calreticulin (CRT) along with inflammatory cytokines and chemokines like CXCL10 and CCL5. Tumor lysis also releases tumor and viral antigens. 3. Dendritic cells infiltrate the tumor using the CCL5 gradient where they bind cytokines, PAMPs/DAMPs, and uptake tumor antigen. Binding of these stimulatory molecules enhances DC maturation and differentiation leading to higher expression of MHCs, CD80, CD86, CD40 and IL12. 4. BATF3+ CD103+ cross-presenting DCs move to the lymph node using CCR7 on their surface to follow the CCL21 gradient and activate tumor and viral antigen specific T cells. 5. Newly stimulated T cells traffic back to the tumor using surface CXCR3 to follow the CXCL10 gradient. 6. Tumor specific T cells recognize and kill tumor cells, leading to tumor clearance

Innate Immunity and Viral Sensing of OVs

When a virus enters and replicates within a cell it is sensed by DNA or RNA sensing machinery such as cGAS/STING and RIG-I ² (Fig. 2). However, expression of viral sensing machinery is often altered in tumors. For example, STING has been found to decrease in expression with increased stage in colon adenocarcinoma and malignant melanoma samples, however it is increased in head and neck squamous cell carcinoma compared to normal tissue³. Expression of RIG-I is also reduced in tumors through both epigenetic alterations⁴ and targeted degradation⁵. However, interest in targeting nucleic acid sensing mechanisms has expanded with the rise of immunotherapy as a way to initiate inflammatory cell death of tumor cells to break tumor tolerance^6,7. While previous research focused on avoiding innate sensing to increase viral replication, more recent data suggests that triggering innate sensing mechanisms may be critical to generate a robust anti-tumor response.

Figure 2: Innate sensing of OVs leads to induction of ISGs and cytokines.

Both RNA and DNA viruses bind the tumor cell surface, using various surface receptors for entry, and enter the cell using receptor-mediated endocytosis. The decrease in pH in the vesicle leads to viral exit into the cytosol. RNA viruses replicate in the cytosol where they are detected by RIG-I or MDA-5. Viral RNA binding to RIG-I/MDA-5 triggers N-terminus caspase activation and recruitment domains (CARDs) to bind CARDs present on mitochondrial antiviral signaling protein (MAVS) at the surface of mitochondria. This leads to the phosphorylation of IRF3, IRF7, and NFkβ (through activation of TBK1 and IKKe). DNA viruses are detected by cGAS, which is found in the plasma membrane, in the cytosol. The binding of cytosolic DNA to cGAS leads to its activation and production of cGAMP from GTP and ATP which then binds STING on the surface of the ER. Activated STING activates IRF3 and NFkβ through TBK1. Activated NFkβ, IRF3, and IRF7 translocate to the nucleus downstream of STING or RIG-I/MDA5 activation and induce the transcription of a host of genes. These include type I interferons, ISGs, IL6, and TNFα. Activation of cGAS/STING and RIG-I/MDA5 can also increase immunogenic cell death, leading to release of PAMPs and DAMPs like calreticulin, heat shock proteins, ATP, and HMGB1. Type I IFNs, stimulatory cytokines, and PAMPs/DAMPs enhance the maturation and antigen uptake of dendritic cells leading to better T cell priming.

DNA Sensing

When dsDNA is found in the cytosol, such as in the event of a viral infection like vaccinia virus (VV) or herpes simplex virus (HSV), cGAS binds the DNA and then dimerizes and produces cGAMP⁶ (Fig. 2). cGAMP binds to STING at the surface of the ER leading to conformational changes that recruit TBK1, which phosphorylates STING and recruits and phosphorylates IRF3. Phosphorylated IRF3 can enter the nucleus and induce gene expression of type I interferons and interferon stimulated genes (ISGs). In the context of oncolytic viruses, there has been interest in reducing the ability of STING to identify viral DNA as it can lead to rapid viral clearance. In the development of oncolytic HSV1, the viral gene γi34.5 was deleted as it is critical in neurovirulence⁸. The protein product of γi34.5, ICP34.5, directly binds STING and inhibits its activity, leading to a reduced STING response to oncolytic HSV and increased viral replication⁸. While this deletion is critical for the safety profile of HSV, this may have detrimental effects on the activation of the adaptive immune system. STING activation can induce strong antitumor immunity in an APC and CD8+ T cell-dependent manner^9,10. More recently, the importance of STING activation in the context of OV therapy has been directly shown. STING knockout lines of LLC and MC38 were found to have increased viral replication and reduced viral lysis when treated with oncolytic HSV¹¹. However, STING knockout tumors had reduced immunogenic cell death in response to HSV infection by measure of HMBG1 and ATP release, and reduced IFNβ, CXCL10, and CCL5 production. These cytokines and chemokines are important for the activation and recruitment of dendritic cells (DCs) and CD8+ T cells into the tumor¹². The loss of STING in these models ultimately led to a loss of responsiveness to HSV + αPD1 combination therapy¹¹. This study directly shows the importance of intact STING signaling for a therapeutic response to OV therapy. Another OV, vaccinia virus, does not properly activate STING. While the modified vaccinia virus ankara (MVA) strain of VV can activate STING, the Copenhagen and Western Reserve strains, which are commonly used to generate oncolytic VVs, do not¹³. MVA infection leads to increased IFNβ and CXCL10 production, however both Copenhagen and Western Reserve strains inhibit IRF3 activation and STING dimerization. VV can also inhibit STING activation through the protein B8R, which inhibits the production of cGAMP¹⁴. Knocking this protein out led to increased phosphorylation of IRF3, increased IFNβ production, and an improvement in tumor clearance and survival in the B16 melanoma model, again showing that increasing STING activation can improve therapeutic response¹⁵.

RNA sensing

RIG-I and MDA-5 are pattern recognition receptors (PRRs) used by cells to detect viral and bacterial material as well as cellular damage¹⁶. RIG-I binds to multiple types of foreign double stranded RNA in the cytosol¹⁷(Fig. 2). Binding leads to conformational changes that result in binding of the CARD domains on RIG-I to MAVS at the surface of mitochondria. This leads to activation of TBK1 and IKKe, dimerization of IRF3 and IRF7, and finally expression of type I interferons and other anti-viral genes. Much like STING, RIG-I has recently become of interest in cancer immunotherapy for its use in activating an anti-tumor immune response¹⁸. Vesicular stomatitis virus (VSV) is a negative stranded RNA virus which has been utilized as an oncolytic virus for over a decade¹⁹. Similar to what was seen with STING, when RIG-I was knocked down in tumor lines, expression of IFNβ, IL6, and ISG15 after VSV infection was lost, while viral replication was increased²⁰. Interestingly, unlike with STING loss, RIG-I knockdown LLC tumors had a modest reduction in tumor growth after VSV treatment compared to control LLC. No further analysis of the TIL was performed to determine changes to adaptive immune induction and animals were not followed out past day 10 to determine if this was a sustained reduction in growth. It is likely that the initial boost in viral replication with RIG-I loss led to increased tumor lysis and slowing in tumor growth, however with less adaptive recruitment these tumors may grow out at the same rate as control. Similar findings were observed in an oncolytic poliovirus with knockdown of MDA-5. Infection of melanoma lines with oncolytic poliovirus induced IFNα and IFNβ production, however with MDA-5 knock down phosphorylation of STAT1 and interferon production were lost while viral replication was significantly increased²¹. While phosphorylation of STAT1 is a major mediator of innate immune activation, overall STAT1 expression is reduced in melanoma patients with a worse prognosis due to high methylation in the STAT1 promoter²². Treating the B16 melanoma model with a combination of HDAC inhibitors to reduce STAT1 methylation and reovirus increased STAT1, RIG-I, and MDA-5 compared to virus alone. As this was not tested in vivo it is unknown if this would improve the antitumor response, however this is a promising mechanism to target tumors with reduced STAT1 expression that needs further study.

The importance of innate viral sensing for adaptive immunity

Both STING and RIG-I lead to expression of type I interferons and ISGs which are critical for increased activation of both innate and adaptive immunity. Their loss may increase viral replication; however this comes at the cost of a reduction in the stimulatory PAMPs, DAMPs, cytokines and chemokines that lead to robust adaptive recruitment to the tumor. These studies show the importance of viral sensing to the development of a complete anti-tumor response in OV therapy. Importantly, viral sensing can be increased through proper selection of treatment candidates based on tumoral PRR expression, combinatorial therapy to increase PRR sensing, viral vector selection, and modification of the viral vector. The main benefit of decreased viral sensing is increased viral replication in the tumor, however viral replication may not be required for OV efficacy. Data from multiple groups now show that inactivated or replication deficient viruses are just as or more effective than replicating OVs, by activating innate sensing mechanisms without viral replication^23–27. While overall the adverse events (AEs) from OVs are more mild than other cancer treatments²⁸, the most common severe AEs (>3) are fatigue, fever, leukopenia and neutropenia²⁹, which can all occur because of uncontrolled viral replication, an especially important point to keep in mind as not every OV has available antivirals and many cancer patients are already in a weakened state. Stimulating innate sensing while reducing viral replication to maintain adaptive immune stimulation may represent a useful path forward in OV therapy.

Activation of adaptive immunity

ICD and adaptive activation

The job of the innate immune system is to recognize and control foreign invaders and trigger the adaptive immune response. In the context of a tumor treated with an OV, the “invader” is the OV and the ability of the virus to stimulate the innate response through PRRs and ICD may directly affect the magnitude of adaptive response generated. Different oncolytic viruses induce different types of ICD in tumor cells; adenovirus, a dsDNA virus, induces autophagy, necroptosis, and pyroptosis, while VV induces mostly necroptosis³⁰. These forms of ICD are increased by activation of cGAS/STING^31,32,33 and RIG-I.^34,35 Importantly, ICD is an important aspect of the oncolytic mechanism³⁶ as it leads to the release of PAMPs and DAMPs and type I IFNs that increase DC maturation, chemokines that recruit DC and T cells to the tumor, and increase antigen uptake by antigen presenting cells^16,37,38. In fact, generating an ICD enhanced version of VV (FUVAC), led to a significant increase in CD8+ T cell infiltration to the tumor and an abscopal response in untreated tumor lesions³⁹.

Type I IFNs and BATF3+ DCs are critical for OV therapy

Type I interferons generated by PRR sensing act on BATF3⁺ DCs to enhance their activation and maturation. BATF3⁺ DCs are critical to immunotherapy as they cross-present tumor antigen to Th1 cells⁴⁰ and produce IL12, a critical cytokine for Th1 induction⁴¹. Treating mice bearing B cell lymphoma tumors with Newcastle Disease Virus (NDV) in combination with Flt3L, which increases DC proliferation, resulted in increased DC activation compared to NDV alone, 100% of mice undergoing a complete response, and fewer metastases in rechallenge⁴². Arming VV to express DNA-depending activator of interferon-regulatory factors (DAI-VV), which is a cytosolic DNA sensor that can activate IRF3, also led to upregulated expression of dendritic cell maturation markers and PRRs⁴³. DAI-VV treatment yielded increased CD8+ T cell infiltration and better tumor growth control in B16-OVA tumors compared to control VV. Interestingly, heat inactivated VV was also found to improve DC maturation through increased IFNβ and CCL5 production by tumor cells as well as increased cGAS-STING activation in DCs compared to live VV²³. Treatment with inactive VV resulted in more tumor regressions in B16 and MC38 than live VV. In all of these examples, treatment in BATF3−/− mice resulted in a loss of treatment benefit^{23,24,42–44}. Treatment benefit was also inhibited by treatment with αCD8 depleting antibodies⁴², αIFNAR⁴² blocking antibodies, or by using IFNAR−/− mice⁴⁴. In the inactivated virus studies, the therapeutic effect was lost when cGAS-STING expression was lost specifically in DCs^23,24. These studies demonstrate the critical role in OV treatment efficacy of complete maturation of DC cells and the role that type I IFNs and cGAS-STING play in it. Importantly, this demonstrates that increasing DC maturation, specifically by targeting BATF3⁺ DCs via cGAS-STING may be a viable combinatorial pathway to increase treatment efficacy.

Virus-elicited adaptive immunity is both tumor and virus specific

It is well documented that CD8 T cell depletion results in a loss of therapeutic response to oncolytic viruses^{15,42,44–50} and as such it is now widely accepted that the CD8+ T cell response is critical to OV therapy. While there is much data suggesting that pre-existing virus-specific immunity is detrimental due to depleting antibodies, there are now data showing that preexisting immunity may enhance the immune response. Mice that received prior immunization to NDV had superior tumor clearance and survival compared to mice without ⁵¹. Expectedly, these mice did have reduced viral replication, however this was not a detriment to efficacy as was expected. In another study, anti-tumor immunity was enhanced by pre-immunizing mice to tetanus, then, after B16-OVA implantation, were treated with an oncolytic adenovirus designed to act as a booster to both tetanus toxin and tumor antigen⁵². In a clinical trial, multiple myeloma patients treated with oncolytic measles virus who underwent a complete response were found to have pre-existing T cells specific to measles virus proteins⁵³. While this was a small-scale study it helps to demonstrate human relevance to pre-clinical data and that this strategy may boost response to OVs and needs further investigation.

Another characteristic of OVs is the ability of the OV itself to reverse the inhibition of antigen presentation within the tumors themselves⁵⁴. This, combined with release of tumor antigen as tumor cells are lysed, and increased DC maturation leads to increased TCR diversity in treatment-responsive tumors. In a pre-clinical lung model, adenovirus treatment led to tumor infiltration of T cells that responded to multiple predicted tumor epitopes⁵⁵. Interestingly, giving intratumoral CpG or poly IC alone, which stimulate cGas-STING and TLR3 respectively, did not result in this same epitope spreading while adding αPD1 in combination with adenovirus increased the effect. In the MC38 model VV that expressed GM-CSF and αPDL1 in MC38 increased mature DCs and CD8+ infiltration into the tumor⁴⁹. These tumor-infiltrating CD8s were tumor neoantigen specific and produced more cytokines than control VV-generated CD8s. This phenomenon has also been observed in patients. In a Phase 1b study, long term survivor PDAC patients treated with reovirus, pembrolizumab, and chemotherapy, had increased TCR clonality in the periphery, of which the majority of expanded clones were new clones generated post-treatment, compared to short term survivors ⁵⁶. Patients with metastatic melanoma or liver cancers treated with PexaVec (an oncolytic VV) had increased tumor antigen specific T cells⁵⁷. Patients with tumor necrosis, often associated with poor prognosis, were found to have significantly reduced T cell diversity post-treatment compared to patients without necrosis, while patients without tumor necrosis had increased diversity after treatment, suggesting that an increased TCR diversity, suggesting epitope spreading, is beneficial. Patients with primary cutaneous B cell lymphomas treated with TVEC had expansion of clonal and unique CD8+ T cells in the injected lesions while the non-injected lesions saw expansion of CD4+ T cell unique clones⁵⁸. Clonal CD8+ T cells had higher expression of effector genes such as perforin and granzymes compared to CD8+ T cells with unique clones. These human data suggest that when treatment is beneficial there is both an increase in TCR diversity, possibly due to epitope spreading, as well as clonal expansion, likely of major tumor or viral epitopes found in the tumor. Importantly, this is directly linked to the quality of the DC maturation that occurs during the initial phases of OV treatment.

Improving oncolytic viral therapies through treatment combinations and viral engineering

While TVEC was approved by the FDA for use in the US in 2014, there have been few additional approvals globally in the subsequent decade, despite significant interest in improving OV therapies. One of which is improving the methods of delivery of the virus itself, but as this has been extensively reviewed elsewhere we will not cover it here^59–62. Much OV research now focuses on combinatorial therapies, both in administering separate therapies in a determined sequence or engineering the virus to produce additional therapeutics. In this section, we will outline the various combination therapies in preclinical and clinical studies and how they interact with what we suggest as a putative “OV Immunity Cycle”: the interaction of OV therapy and the tumor immunity cycle (Fig. 3).

Figure 3: Targeting the OV immunity cycle.

Interventions are listed at each step of the OV immunity cycle that can increase the efficacy of OV therapy.

Targeting innate sensing, ICD, and DCs

Recently, targeting innate sensing mechanisms as an adjuvant to immunotherapy has garnered increased interest. Many of these are now being combined with OVs to improve response. One way to accomplish this is to target the signaling molecules downstream of PRRs. Preclinical research show that combining JAK/NFkB inhibitors to decrease transcription of ISGs and other STING-targets or AKT/mTORC inhibitors to inhibit ISG and type I IFN translation both yield increased viral replication^63,64. However, it is important to weigh the cost of inhibiting IFN and ISG expression as many studies have demonstrated the importance of these factors in stimulating the adaptive immune response during OV treatment. To this end, more studies are needed, especially utilizing immunocompetent mouse models to evaluate the effects of these types of inhibitors on OV therapy.

STING agonism has been studied extensively as an adjuvant to other types of immunotherapy, but also in its own right. There are many preclinical and clinical trials ongoing with these agonists^6,7. As we have discussed in this review, STING activation is critical for full activation of the adaptive immune response, however, many tumors either express low levels of STING³ or the virus itself may evade STING detection^13,14. This suggests that STING agonism is a logical combination with OV therapy. This may be achieved through a variety of mechanisms: engineering the virus to stimulate STING, engineering the virus to express a STING agonist, selecting a viral vector that better stimulates STING, selecting tumors that express higher levels of STING. While more preclinical data is needed to validate this strategy, the existing data are promising.

RIG-I agonists are also in development, however, are not as widely studied as STING for immunotherapy. Agonism of RIG-I has been shown to reduce tumor growth and extend survival in the MC38 and YUMMR models when delivered intratumorally ⁶⁵ as well as increase adaptive immune activity when given as an adjuvant to the influenza vaccine ⁶⁶. While it is not currently being tested in combination with OVs, this may be an innate mechanism to target when using RNA based viruses.

While STING and RIG-I agonism can both increase ICD^31–35, there are other ways to increase ICD being studied in combination with OVs as well. A VV that delivered MLKL, an effector of necroptosis, improved the anti-tumor response in B16 tumors compared to control VV and increased levels of IFNγ in treated tumors⁶⁷. In another study, an AdV that induced ICD combined with αPD1 reduced primary tumor growth and metastases in CT26 and PAN02 models⁶⁸. This study did not compare to a virus that did not induce ICD however.

Tumors have varying expression of innate sensors and their signaling molecules and viruses can evade detection. As a result, there are many avenues which can be exploited to increase innate sensing and fully stimulate an adaptive anti-tumor response.

Targeting trafficking and antigen presentation/recognition with CAR-T cells

OVs have also been used in combination with CAR-T cell therapy as a method to alter the TME from “cold” to “hot” or to enhance CAR-T cell therapy through delivery of other modulators into the TME⁶⁹. As one major limitation of CAR-T cells is the lack of surface antigen on solid tumors, VV was engineered to deliver a truncated CD19 into solid tumors for surface expression and allow for targeting with CD19 CAR-T cells⁷⁰. Treatment with this VV and CAR-T cells improved response over either treatment alone in both a xenograft and immunocompetent MC38 model. Another major limitation of CAR-T cell therapy in solid tumors is CAR-T cell entry. Evgin et al showed that while pre-treating with VSV that produces IFNβ (VSV-IFNβ) prior to EGFR-CAR delivery induced chemokine changes that should increase tumor infiltration, the VSV-induced alterations actually led to increased apoptosis of the CARs⁷¹. In a follow-up study, they found that giving VSV-IFNβ five days after CAR-T cell administration resulted in an expanded dual-specific CAR-T cell population in the tumor which recognized both EGFR and a VSV epitope⁷². These cells produced more cytokine and acquired an effector memory phenotype at a higher frequency than CAR-T cells from mice treated with PBS. These dual-specific CAR-T cells could be reactivated with a second “boost” of the virus. This was also shown to work with reovirus instead of VSV. Most interestingly, this report suggests that with the correct scheduling of CAR-T and virus administration, the viral infection itself may provide sufficient milieu alongside cognate TCR antigen to enhance effector function and maintain long term circulation of the CARs, unlike in other models where native TCR engagement has been shown to be detrimental to CAR-T cell survival⁷³. These kinds of studies are critically important as there are ongoing trials combining CAR-T cells and OVs (NCT03740256, NCT01953900). Combining CAR-T cells and OVs is a promising strategy to overcome issues in antigen expression and recognition that inhibit the success of both therapies.

Targeting T cell trafficking, infiltration, and tumor cell killing

One of the major benefits to using an oncolytic virus is the ability to engineer it to deliver cargo directly into the tumor microenvironment. As the virus replicates it will translate any genes that have been added into its genome and produce essentially anything you want; checkpoint blockade antibodies, small molecule inhibitors, cytokines and chemokines are all some examples that have been used. Commonly, this method has been used to increase the immune stimulating abilities of the OV; FDA approved T-Vec and clinical trial candidate PexaVec both express GM-CSF. This strategy is now being utilized in glioblastoma which grows in a classically immune-privileged site. An oncolytic HSV1 strain modified to express cetuximab, an EGFR antibody, and CCL5, the chemokine that recruits immature DCs, heterodimers which block EGFR signaling while targeting CCL5 expression in the tumor⁴⁵. In an immunocompetent GBM model, this fusion virus significantly increased survival over control HSV alone and control HSV with CCL5-cetuximab fusion given by osmotic pump. The fusion expressing virus led to significant increases in NK cell, phagocytic macrophage, and IFNγ and granzyme B producing CD8+ T cells in the tumor compared to control virus. Another mechanism to increase infiltration and tumor cell killing is to engineer the OV to express bispecific T cell engagers, or BiTEs⁷⁴. BiTES are bispecific antibodies designed to engage both antigen on tumor cells and a surface molecule on T cells to bring the two cell types together and induce tumor cell killing. The clinically approved BiTE, blinatumomab, which targets CD19 and CD3, works well in acute lymphoblastic leukemia (ALL) but needs constant infusion, has high toxicity, and has difficulty penetrating solid tumors such as Non-Hodgkin Lymphomas. To address this, an oncolytic VV was engineered to produce the BiTE and tested in NSG mice bearing Raji tumors with adoptively transferred T cells⁷⁵. The fusion virus led to increased T cell infiltration, tumor clearance, and survival compared to control virus or blinatumomab alone. In another study, VV was engineered to express an EpCAM/CD3 BiTE which led to increased T cell infiltration, activation, and tumor clearance in EpCAM⁺ tumor models but not EpCAM⁻⁷⁶. Similar findings were reported using an adenovirus to deliver a MUC1/CD3 BiTE in humanized mouse models⁷⁷. These fusion viruses can successfully enhance T cell infiltration directly where it is needed to enhance therapy efficacy.

Concluding Remarks

Traditionally, OVs have either been armed or combined with immune stimulating agents, such as GM-CSF, IL2, or αPD1. However, this has not led to the clinical successes observed by checkpoint blockade alone in many tumor types⁷⁸. How can we utilize our understanding of the OV mechanism of action to design better therapies and better combinations? This includes better selection of possible responding patients and better selection of viral vectors to give the best chance of success (see Outstanding Questions).

Outstanding Questions Box.

• Important questions for future research should be summarized in a box (not included in box count or element limit). This is an excellent opportunity to offer input and guidance on new directions for the field.

  • ‐
    What are the most important suppressive mechanisms that prevent OV therapy from working? Can we engineer the virus to deliver modulators to reverse that suppression?
  • ‐
    What is the effect of pre-existing immunity on OV therapy in patients?
  • ‐
    Can selecting patients based on tumoral expression of STING/STAT1 improve response to OVs?
  • ‐
    What is the effect of JAK/STAT or AKT inhibition to increase viral replication on tumor growth in immunocompetent models? What is the requirement of viral replication in OV therapy in patients/immunocompetent mouse models?
  • ‐
    How can we match viral characteristics to those of a patients tumor (immune infiltration, PRR expression, antigen load/mutational burden, suppressive mechanisms etc) to improve response?

First, we now better understand the importance of PRR signaling in the generation of a complete and robust adaptive immune response. However, many tumors either express PRRs at low levels or we have selected/engineered oncolytics to evade PRR binding. One way to improve OV therapy would be to enhance this signaling. This could be achieved by selecting patients based on cGAS-STING/RIG-I expression in their tumors, using virus backbones that can be detected by PRRs, engineering OVs to increase PRR stimulation through expression of small molecules, or utilizing heat-inactivated forms of OVs which better stimulated STING than live versions of the same virus. Second, we also understand that the maturation of DCs is critically important for generating an expanded CD8+ response. While improving PRR stimulation can improve DC maturation, treating with DC agonists like Flt3L or using the virus to deliver chemokines into the tumor where they can bring DCs and CD8s into the tumor, are both ways to improve this part of the response. Combining with therapies that enhance or induce expression of specific antigens may also help. Finally, once the new infiltrate reaches the tumor it must contend with the suppressive TME. More research must be done on the suppressive mechanisms that impede OV therapy so that they can directly be targeted. Our group recently published work that shows that in VV resistance, high TGFβ maintains the suppressive capacity of tolerogenic populations like regulatory T cells after VV treatment⁷⁹. By engineering VV to express a TGFβ inhibitor we directly target immunosuppression and not only enable VV therapy, but also synergize with other therapies like checkpoint blockade.

Oncolytic virotherapy is a complicated and multistep therapeutic which requires almost every aspect of the immune system to generate an anti-tumor response. Achieving therapeutic efficacy may involve sequencing with other therapies as well as better viral engineering. We posit that the field is just beginning to scratch the surface of what can be accomplished with engineered, oncolytic viruses, especially in terms of payload selection. So far, OV biologists have taken a ‘kitchen sink’ approach to transgene selection, adding as many immune stimulators as one can fit into the viral genomes. However, we believe the genetic payload aspect of OV therapy is an incredible opportunity for engineering, beyond stimulatory cytokines or PD-1 blockers. Viruses can be designed to work together or even work with other forms of therapy. This is an exciting platform to leverage synthetic biology, in which viruses could be engineered to deliver specific or even orthogonal signals to other cell types, especially engineered cells like CAR-T, for instance. But most importantly, we must improve our understanding of how OVs interact with the tumor microenvironment and the immune system. More fully understanding the immunobiology of OV therapy will propel the design of next-gen viral strategies and deliver the hope of curative responses to patients.

Highlights.

• Highlights are a short collection of bullet point statements (3–5) that concisely convey to the reader the recent advances in the area, including emerging concepts and/or distinctions, that constitute a main motivation for the discussion developed in the article.

  • ‐
    Contrary to the long-held belief that anti-viral immunity is a detriment to OV therapy, innate sensing of the virus is required for complete DC activation and anti-tumor immunity
  • ‐
    Type I IFNs and BATF3+ DCs are required for the OV anti-tumor response
  • ‐
    Viral replication may not be required as inactive virus can still activate cGAS-STING, chemokine and cytokine production, and stimulate an efficient anti-tumor immune response
  • ‐
    Patients who respond to OV therapy both generate new TCR clones and have expanded clonal populations
  • ‐
    OVs and combination therapies should be designed to target and enhance these properties

GLOSSARY

Adenovirus (AdV)

a 24–46 kbp dsDNA virus that uses CD46, coxsackie/adenovirus receptor, sialic acid, and MHC for cell entry and replicates in the nucleus

Basic Leucine Zipper ATF-Like Transcription Factor 3 (Batf3)

AP-1 member transcription factor required for cDC1 and CD103 cross-presenting dendritic cell development

cGAMP (2′3′-Cyclic GMP-AMP)

intracellular second messenger synthesized in response to cytosolic dsDNA sensing which activates STING

Cyclic GMP-AMP synthase (cGAS)

enzyme that binds cytosolic dsDNA and generates cGAMP

Cross presenting DCs

DCs that pick up non-self antigen and present it on MHC Class I

Damage associated molecular patterns (DAMPs)

also known as alarmins, are self-molecules released during damage and death that trigger an inflammatory response

Flt3L

a cytokine and growth factor that binds Flt3R to promote proliferation of DCs

Herpes simplex virus (HSV)

152 kbp dsDNA virus that uses host surface glycoproteins for cell entry and replicates in the nucleus

Immunogenic cell death (ICD)

any type of cell death that releases inflammatory molecules, typically intracellular products that should not be found extracellularly, to elicit an immune response

Interferon stimulated genes (ISG)

are genes turned on by interferon or pathogen sensing that produce anti-viral gene products

MDA5

RIG-I like PRR of cytosolic free RNA which triggers an interferon response through IRF3 and IRF7

Newcastle disease virus (NDV)

a 15.2 kb negative sense ssRNA paramyxovirus which binds host surface sialic acid-containing compounds for entry and replicates in the cytosol

Pathogen Associated Molecular Patterns (PAMPs)

microbial molecules with defined “patterns” that trigger host innate sensing mechanisms, such as PRRs or Toll-Like Receptors

Pattern Recognition Receptor (PRR)

cell receptor that recognizes PAMPs and DAMPs and triggers an interferon response

Reovirus

a 24 kbp dsRNA virus that uses host cell surface sialic acid and junctional adhesion molecule (JAM) for cell entry and replicates in the cytosol

RIG-I

a PRR that binds foreign RNA in the cytosol to trigger an interferon response through IRF3, IRF7, and NFkB activation

STING (Stimulator of Interferon Genes)

adaptor protein that binds cGAMP, produced by cGAS in response to cytosolic dsDNA, to trigger the type I IFN response

TCR clonality

a measure of how expanded an identical TCR population is

TCR diversity

a measurement of unique TCRs and their relative abundance within the population

Type I IFNs

interferons which bind IFNAR1 and IFNAR2 to stimulate an anti-viral response (i.e. IFNβ and IFNα)

Vesicular Stomatitis Virus (VSV)

an 11k nucleotide negative sense ssRNA arbovirus that uses LDL receptor for entry and replicates in the cytosol

Vaccinia Virus (VV)

a 190 kbp dsDNA orthopoxvirus that uses surface glycoproteins for host cell entry and replicates in the cytosol