White paper on microbial anti-cancer therapy and prevention

Virotherapy and cancer vaccines

Most virus-based therapies are multi-mechanistic. These therapies function by delivering genes into the cancer host cells. Many of these delivered genes are designed to either directly engage the immune system or to lyse cancer cells. This lysis leads to the release of cancer specific antigens that in turn stimulate the acquired immune system. Virus therapies can be grouped into two types: cancer vaccines and virotherapies. Vaccines encode and express cancer-specific antigens or epitopes. In contrast, virotherapies do not encode exogenous tumor antigens but drive lytic mechanisms that kill cells to unmask or unleash endogenous antigens from the cells that become infected. “Oncolytic vaccines” are a hybrid strategy, in which an oncolytic vector also encodes a tumor antigen or neo-epitope.

Viral therapies can utilize either replicating or non-replicating viruses. Replication-competent viruses retain the ability to replicate and spread to adjacent un-infected tumor cells and are, in theory, dose amplifying. Eventual death of virus-infected tumor cells ensues a second mode of efficacy resulting from host anti-tumor immune responses roused by the release of tumor antigens along with inflammation due to infection. Together this has the potential to educate a host memory response to clear both injected lesions as well as un-infected metastatic disease.

Cancer vaccination: Gene delivery to engage the immune system

Viral vectors are the default selection for gene-delivery therapies. Research over the past three decades has led to a deep understanding of viral tropism, genome sequences, regulatory elements and immune-evasion mechanisms. These efforts have culminated in the development of a versatile arsenal of increasingly tumor-selective viral-vector technologies. Replacement of naturally-occurring viral genetic-regulatory elements with tumor-specific elements has been a key enabling feature that has led to the generation of numerous tumor-selective viral vectors [60, 61]. These modified viruses limit gene expression and infectivity to tumor cells while sparing normal cells, allowing for the incorporation of a wide array of different genes encoding for potent effector proteins such as cytokines, chemokines, protein toxins, gene silencing RNAs, micro RNAs, and prodrug-converting enzymes [62]. Viruses have been used as vaccination agents, either by inducing cross-reactive immunity, or more commonly being engineered to express specific antigens to be targeted by adaptive immune responses.

To enhance the immune response, the processing of specific antigens can be altered. For example, gene sequences encoding fusion proteins linked to intracellular-trafficking elements have been demonstrated to improve and enhance antigen presentation [63, 64]. The potential to locally augment an adaptive immune response to tumor neoantigens is quite promising, as there is a consensus that shared mutated antigens are uncommon, and the majority of mutated antigens serving as neoantigens are private. Consequently, local secretion of a variety of immune-stimulating molecules, such as GM-CSF and IL-12 have been engineered into microbes, as well as adaptor proteins such as MyD88 [65, 66].

Another approach to cancer vaccination is “in situ” vaccination (ISV). Vaccines exploit antigens to be recognized by the immune system and adjuvants to activate the immune system, but for cancer, the problematic issue has been to choose exactly which antigens are the “right” ones. ISV uses the tumor as the source of the antigen and injects the adjuvant directly into the tumor. The overall goal is to generate a strong local antitumor response that reverses the local tumor-mediated immunosuppression and then to have that local response become systemic to fight potential or recognized metastatic disease [67, 68]. For over 30 years, the standard-of-care to treat superficial bladder cancer after surgery has been applying Bacillus Calmette-Guerin to the bladder lumen as an ISV immune adjuvant [69]. Another current ISV treatment in clinical usage is resiquimod, a TLR 7 agonist, for cutaneous T cell lymphoma treatment [70]. There are a variety of ongoing studies investigating the best approaches to in situ vaccination, such as developing a plant virus, cowpea mosaic viral-like nanoparticle, as a potent ISV reagent [71].

Oncolytic viruses: Vaccines and virotherapy

One limitation of cancer vaccines is that, although tumor-associated antigens have been well characterized, the presence of tumor-specific neoantigens generated by somatic mutations within the genome of a cancer cell, are most commonly private, with shared antigens representing a very small percentage of known antigens [72]. Consequently, strategies to identify and generate tumor-specific immune response are being sought. To create immunogenic viral vectors, natural viruses have been modified to express tumor-specific antigens during the tumor-cell-lysis portion of the viral-infection cycle. This modification results in in situ “oncolytic vaccination”, at least partly promoted by antigen cross-presentation [73]. To better promote the immune response against these host antigens, several oncolytic viruses encode for immunostimulatory cytokines such as Type-I interferons or GM-CSF [73]. An alternative approach to strict viral-mediated oncolysis is tumor-selective dissemination and expression of a prodrug-converting enzyme followed by administration of the prodrug (e.g. Toca 511/FC) [74]. There are numerous active clinical trials with oncolytic viruses (Table 2).

One advantage of recombinant viruses that are oncolytic, is that they are capable of causing tumor cytoreduction, but also capable to elicit adaptive immune responses to the antigens, including neoantigens released by tissue injury and death induced by the viral life cycle. Immune activation by oncolytic-virus infection of tumor tissue comprises both, immediate effects of innate immunity and also adaptive responses for long-term tumor-specific activity. Much of the enthusiasm surrounding oncolytic vaccines revolves around their appeal as multi-mechanism therapeutics [21]. Oncolytic viruses are proposed to function through a combination of (1) selective killing via tumor-specific cell lysis (oncolysis) and (2) activation of innate and adaptive immunity via immunogenic cell death, immunostimulation from recognition of viral constituents, and release of both viral and tumor antigens [75]. As mentioned, oncolytic viruses can also be engineered to encode transgenes to further engage the host immune response.

This approach to microbial-based cancer therapy may be particularly well suited for use in indications not considered to be highly immunogenic, but where overexpression of specific tumor-associated antigens could be of value. These recombinant microbes are engineered to deliver or express tumor associated antigens in professional and immature antigen-presenting cells in vivo to break peripheral tolerance of tumor-antigen-specific effector cells. Measurable clinical responses to microbial therapies designed for this purpose have been achieved and, while response rates are low, they are significant, appear to be durable, and highlight the promise of this approach.

Bacterial therapy

Similar to viral therapies, bacterial therapies can function predominantly by direct oncolysis or by engaging the immune system. Direct oncolysis can be mediated by secreting exotoxins or competing for nutrients [76]. Alternately, intracellular bacteria can kill host tumor cells by inducing apoptosis or by uncontrolled proliferation resulting in bursting of the infected tumor cells [77]. In addition, bacterial infections inevitably activate both innate and adaptive arms of the immune system, which target not only the bacteria but also tumor cells [78].

An essential strength of bacterial therapies is their specific targeting of cancerous cells and tissues. Each modality targets in a distinct way. For example, active bacteria have been engineered to only colonize the microenvironment of tumors [79, 80]; oncolytic bacteria have been engineered to induce cell death specifically in cancer [81]; and immune-sensitizing bacteria have been engineered to induce responses to cancer-specific antigens either directly [30] or indirectly through epitope spreading [82]. Because of these varying targeting mechanisms, microbial therapies are well-suited as therapies for metastatic disease.

Immune engagement with bacteria

The intrinsic ability to stimulate the innate immune system makes bacterial-based cancer vaccination strategies an attractive approach. Many genera of bacteria accumulate in the tumor microenvironment and once there can convert the tumor microenvironment (TME) through proinflammatory responses. Tumor-colonizing bacteria alter the immune suppressive environment of tumors and make them immune stimulating. This conversion can be achieved through different approaches, such as (1) incorporation or cloning of adjuvants into the bacteria, (2) cloning highly immunogenic antigens as an alternative for neo-antigens, (3) encoding with immunogenic cytokines which evoke anti-tumor immune responses, (4) cloning of checkpoint inhibitors into the bacteria, or (5) depletion of immune suppressing macrophages or myeloid-derived suppressor cells (MDCS). Species that have been investigated for this approach include attenuated Listeria monocytogenes, Salmonella enterica, and Clostridium novyi-NT. All of these bacteria selectively survive and multiply in the TME. For example, attenuated Listeria monocytogenes have been designed to deliver recombinantly-expressed tumor-specific antigens [84]. Similarly, strains of attenuated Salmonella enterica have been engineered to secrete biologically-active cytokines such as IL-2 [18, 85].

Many classes of microbial anticancer therapies rely upon induction of immunogenic cell death (ICD) to propagate effective anti-tumor immunity. ICD is recognized as a critical process that initiates the tumor-immunity cycle and ultimately leads to an adaptive immune response [86]. In contrast to tolerogenic cell death and other forms of tumor cell demise, ICD is characterized by three distinct molecular events which are necessary to prime and activate tumor-infiltrating dendritic cells: (1) translocation of calreticulin from the endoplasmic reticulum to the cell surface (i.e. ecto-calreticulin), (2) passive extracellular release of high-mobility-group box 1 (HMGB1) from the nucleus, and (3) extracellular release of ATP [87, 88]. These molecular “danger” signals collectively function as damage-associated molecular patterns (DAMPs) which recruit professional antigen-presenting cells into the TME which can then uptake tumor-associated antigens released from dying tumor cells.

Bacterial-based vectors naturally stimulate the innate system through activation of multiple immune pathways. Diversity among membrane components between bacterial species manifests in differential expression and presence of structurally-conserved pathogen-associated molecular patterns (PAMPs). These molecules include flagella, pili, adhesins, lipotechoic acid and lipopolysaccharide and each activates specific toll-like-receptor (TLR) family members to elicit distinct innate immune-signaling cascades that ultimately translate to a comprehensive immune signature unique to each bacterial organism [89]. Different bacteria engage the innate immune system in distinct ways including repolarization of tumor associated macrophages (TAMs) from an immunosuppressive pro-tumor M2 phenotype to a M1 anti-tumor phenotype; selective elimination of both peripheral and tumor-associated myeloid derived suppressor cells (MDSCs); and promotion of dendritic-cell maturation at the tumor site [90]. CD8⁺ T cells are a major player in the adaptive antitumor immune response [26, 91]. A bacterial infection establishes an inflammatory environment that sensitizes CD8⁺ T cells to low-density tumor antigens by enhancing proximal T-cell receptor (TCR) signaling [92].

Direct tumor destruction with bacteria

In addition to their intrinsic anti-cancer effects, bacterial vectors can be genetically engineered or modified to express proteins with direct tumoricidal properties [18]. Because of their ability to host large amount of foreign DNA, bacteria can be used as a platform for synthetic biology and the construction of entire functional genetic circuits [93]. Bacteria have the capacity to accommodate heterologous DNA of larger than 300 kb [94]. The possibility of delivering ubiquitously-toxic molecules relies on the bacterial ability to selectively colonize, proliferate and express genes only in tumors [95]. This specificity is maintained by introducing stable genetic circuits that are selectively responsive to unique qualities of the TME, such as differential pH, nutrient and/or oxygen availability [96, 97]. To further limit toxicity, quorum-sensing systems have been developed that focus expression to tumors and prevent expression in normal tissue [98].

The most prominent bacterial toxins exploited as therapeutics are pore-forming cytolysins derived from Escherichia coli [99, 100], or Staphylococcus aureus [101, 102]. These molecules are natively expressed by bacteria, easily secreted, and induce apoptosis in mammalian cells [102–104]. They are secreted as monomers that form multimeric pores after incorporation into the membranes of eukaryotic cells [105–107]. Several cytotoxic cytokines that have been explored are FAS ligand (FASL), TNF-related apoptosis-inducing ligand (TRAIL) and TNFα [25, 108–111]. In addition to being immunogenic, these molecules directly induce apoptosis [25] and are more toxic to cancer cells than normal cells [25, 108]. As therapeutics, TNFα and TRAIL are effective against bladder, breast, colon, glial, lung, ovarian, pancreatic, prostate, and renal cancer cells [25, 112]. However, these cytokines cannot be administered systemically because they would induce significant toxicity. Localized production by bacteria colonized within tumors obviates this problem and focuses treatment on the cancerous lesion.

Immune-cell-targeting bacteria, such as Listeria monocytogenes, can reduce tumor mass by delivering radioactivity. For example, 188-rhenium has been coupled to Listeria through anti-Listeria antibodies and shown to be effective against mice with pancreatic cancer [113, 114]. Bacteria can also be engineered to deliver siRNA. To achieve this strategy, E. coli has been transformed to transcribe shRNAs from a plasmid containing the invasin gene inv and the listeriolysin O gene hlyA. These two genes encode bacterial factors that are needed for transfer of shRNAs into tumor xenografts [115].

Combination of viruses and bacteria with other modalities

The greatest therapeutic strides with oncolytic platforms are likely to be achieved for diverse patients through multipronged pharmacological and immunomodulatory approaches. One approach is to identify chemotherapy drugs that can alter cellular signaling to sensitize tumor cells to viral replication and/or burst. Cellular DNA damage, after induction by radiation or chemotherapy, can augment HSV-1 replication by amplifying viral burst [116]. Based on these observations, G207 has been evaluated for safety and efficacy in combination with radiation for adult and pediatric patients with brain tumors [117]. Another rationale for utilizing high dose chemotherapy has been to use the lympho-depletive effects to counter innate immune cell mediated virus clearance. Encouraging preclinical results from this approach have resulted in a clinical trial (NCT00450814) wherein safety and efficacy of measles virus-NIS has been tested as a stand-alone therapy and in combination with cyclophosphamide. While chemotherapy is known to have immune suppressive effects, manipulating the timing of chemotherapy with viral agents can also exploit the development of anti-tumor immunity due to the release of chemotherapy-induced lymphopenia during the time of infection and hence tumor antigen presentation. This allows for expansion of tumor-selective T cells to maximize antitumor benefits [118, 119].

The greatest attention has been paid to combination with checkpoint-inhibitory antibodies [120–124]. The recent success of immune-checkpoint blockade with humanized antibodies blocking negative regulatory receptors such as CTLA-4 and PD-1 has brought forth renewed interest in enhancing virus mediated anti-tumor immunity in conjunction with these agents. A randomized open Phase II study evaluating the combination of talimogene laherparepvec with ipilimumab (a humanized anti-CTLA-4 antibody therapy) in patients with advanced melanoma revealed a greater antitumor activity of the combination without increasing adverse events [125]. The combination of T-VEC and anti-PD-1 immunotherapy modulates the TME and promotes intratumoral T-cell infiltration resulting in overall and complete responses among patients with advanced melanoma [126]. Oncolytic viruses have been shown to recruit and provide cytokine support to CAR-T cells [127–129]. Many of these strategies have demonstrated a favorable toxicity profile, high overall-response rates, and marked intra-tumoral T cell infiltration and anti-tumor T cell responses [121, 126].

While many bacterial-based therapeutics impart significant single agent activity, combination with other modalities may increase efficacy. Bacterial proliferation is often specific to the immune-privileged necrotic/hypoxic regions of solid tumors [130, 131], sparing the well-perfused regions that will eventually grow back. Traditional cytotoxic therapies, such as chemotherapy and radiation therapy, are most effective against cells in those well-perfused regions. Combining bacteria with these cytotoxic therapies generates synergistic effects [24, 132, 133]. Another strategy employed in combination with bacteria is the expression of cleaving enzymes that activate prodrugs specifically in the TME [134]. For example, delivery of HSV-TK with Salmonella converts gancyclovir to its toxic form and results in a dose-dependent reduction in tumor burden [135]. Similarly, Listeria has been used to deliver prodrug-converting enzymes like purine-nucleoside phosphorylase and a fusion protein consisting of yeast cytosine deaminase and uracil phosphoribosyl transferase [136].

Non-living bacterial strategies

Non-living bacteria-based therapies are alternative strategies that have several advantages over living organisms. This category of therapies includes bacterial minicells, cell-wall complexes, outer-membrane vesicles, and other bacterial derived therapies [137–139]. Bacterial minicells are small chromosomeless bacterial particles that contain all the natural and recombinant components of their parent cells yet are unable to divide and are non-infectious. In addition to being as amenable to recombinant engineering as their parental progenitors (i.e. can be engineered to carry oncolytic proteins, for example), minicells have limited metabolic activity and can be stably loaded with small-molecule cytotoxic drugs by passive diffusion.

Site-specific immunomodutors (SSIs) are composed of components of specific bacterial pathogens. When administered subcutaneously, SSIs result in a recruitment and activation of innate immune cells (including macrophages and NK cells) in the organ in which the bacterial pathogens, from which they are derived, commonly causes infection. For example, QBKPN, derived from a lung pathogen (Klebsiella), stimulates a lung-specific anti-cancer immune response, including macrophage recruitment and M1 polarization, NK cell recruitment and activation, and upregulation of the NKG2D pathway [140]. QBKPN reduces lung-cancer burden in mouse models and downregulates PD-1 and PD-L1 systemically in lung-cancer patients [140].

Prevention with microbial therapies

Microbial therapies have the potential to prevent cancer by several distinct dimensions. First, microbiota can be manipulated to destroy tissue, which can be utilized as a form of cancer prevention. Microbial treatment of a neoplastic lesion may serve not only to eradicate the lesion, but also stimulate an immune response. That response may, in turn, prevent subsequent invasive cancer. Cancer prevention by tissue ablation forms the basis for many contemporary cancer prevention strategies targeting particularly high-risk individuals, including prophylactic mastectomy, oophorectomy, colectomy, and thyroidectomy for significant reduction in the incidence of subsequent breast cancer, ovarian cancer, colon, and thyroid cancer, respectively [141]. A second preventative approach would be manipulation of the microbial pathogens that drive tumorigenesis and are associated with an altered composition of commensal microbiota (dysbiosis). Microbes are estimated to be involved in 15% to 20% of cancer cases. Preclinical studies demonstrate that modulation of inflammation, DNA damage, and metabolite production are potential mechanisms of oncogenesis or tumor suppression, which may be altered with changing the microbial composition [142]. A third approach would utilize microbial prophylactic vaccines to target cancers with viral etiology, similar to HPV and HBV vaccines [143–145]. A fourth strategy would be enhancement of the immune system response by modifying dendritic cells (DCs) to improve vaccine potency.

Lessons from trials with oncolytic viral therapies

In the last two decades, numerous clinical trials have been initiated with oncolytic viruses. These trials have taught many lessons about the mechanisms of cancer vaccination and how to select the most appropriate patients (see the Mayo Clinic Experience with Virotherapy, Case Box 5). For example, experience with cancer vaccines has demonstrated that targeting tumor-specific antigens alone is likely insufficient to generate effective anti-tumor immunity and to prevent immune escape. Optimal T-cell responses come from not only the engagement of the T-cell receptor binding to the MHC-peptide complex (signal 1) but also to a second signal with cytokine stimulation or a costimulatory molecule (signal 2). Two FDA-approved cancer vaccines highlight this concept. Sipuleucel-T (Provenge), an autologous dendritic cell vaccine contains a fusion protein (PA2024) consisting of the target antigen (prostatic acid phosphatase) as well as immunostimulatory GM-CSF, which can augment dendritic cell activity and maturation. Similarly, T-VEC produces GM-CSF to enhance immunogenicity [146].

An intriguing story has emerged suggesting that effective cancer immunotherapy can promote “antigen spread” [137]. The notion of antigen spread suggests that immunogenic cell death of target antigen-expressing tumor cells release non-targeted antigens that prime subsequent adaptive immune responses against non-targeted antigens. This phenomenon has been observed among patients treated with PROSTVAC, which is a prostate cancer vaccine that contains transgenes for prostate-specific antigen (PSA) and three T-cell co-stimulatory transgenes: B7-1 (CD80), leukocyte function-associated antigen-3 (LFA-3), and intracellular-adhesion molecule-1 (ICAM-1) [138, 139]. This observation suggests that T-cells specific for tumor-antigens not present in the initial vaccine construct may be an important additional mechanism of action and potentially indicative of a favorable immune response.

Numerous studies investigating the role of innate anti-viral defense responses in ocolytic viral therapy have uncovered that rapid virus clearance minimizes the initial burst of tumor cells and thus limits virotherapy by negatively regulating virus lytic tumor destruction [147]. While infection-induced host responses can clear the virus, this initial inflammatory response to viral infection is also thought to be an essential step leading to the influx of antigen presenting cells and the development of subsequent anti-tumor immunity [148]. The Yin-Yang balance between innate virus clearance limiting therapeutic index and the stimulation of host anti-tumor immunity has emerged as a heavily debated subject, wherein host innate-defense responses are touted as being both beneficial and detrimental to therapy. Most likely the situation is further compounded by variables that impact host response such as the virus platforms used, different tumor types and stages, TME, and prior-treatment status of the patient.

Lessons from trials with bacterial therapies

Because of the potential toxicity of these therapies, knowing when and how to intervene are critical issues which will have direct impact on clinical outcomes. Dose determination is another important issue with the clinical development of bacterial therapy. Dose escalation is a standard practice in Phase I clinical trials. However, the effective dose for live bacteria depends more on the target tumor tissue where they can multiply than on the administered dose. A large tumor with extensive necrosis and minimal infiltration of inflammatory cells is more likely to support a robust bacterial infection, resulting in pronounced tumor response and toxicity, regardless of the administered dose. In contrast, a small tumor or a tumor with minimal hypoxic/necrotic regions is unlikely to show significant response even if a large number of bacteria are administered.

Immune stimulation and engagement

To date, the complex effects of a viral tumor infection on the TME and the consequences for the tumor-infiltrating immune cell compartment are poorly understood. There is growing evidence that a tumor infection by an oncolytic virus opens up a number of options for further immunomodulating interventions such as systemic chemotherapy, generic immunostimulating strategies, dendritic cell-based vaccines, and antigenic libraries to further support clinical efficacy of oncolytic virotherapy [126]. More specifically, can microbial infection of an immunologically “cold” tumor (i.e. low mutational load and/or nonresponsive to immune checkpoint inhibitor monotherapy) induce conversion into a “hotter” milieu where novel endogenous tumor epitopes are now presented more efficiently to the immune system?

Novel approaches are needed to alter the immunosuppressive microenvironment of tumors and facilitate the recognition of tumor antigens. The lack of immune-cell infiltrates within tumors negatively correlates with the response to immunogenic cell death (ICB) and patient survival [52, 158]. Improved understanding of the critical role and mechanisms of Batf3 dendritic cell antigen uptake and cross-presentation may yield novel insights to optimize dendritic cell-based vaccine strategies. Because innate immune responses against infection and malignancy utilize similar evolutionarily conserved pathways (i.e. cGAS-STING pathway) microbial anticancer agents may be particularly effective at engaging and amplifying anti-tumor immunity [159]. Batf3-dependent CD103⁺/CD8α⁺ dendritic cells (DCs) are critical for cross-presenting tumor antigens in tumor draining lymph nodes (TDLNs). Cross presentation allows for priming and expansion of naïve CD8⁺ T cells that recognize the cognate antigen [52, 160–162]. Possible strategies to expand CD103⁺/CD8α⁺ DCs include systemic administration of FMS-like tyrosine kinase 3 ligand (FLT3L) or intratumoral delivery of immune adjuvants such as TLR3, TLR7, and STING agonists.

Bacterial infections of tumors naturally stimulate an immune response. However, it is not entirely clear how intratumoral infections enhance antitumor immune responses. It has been suggested that the antitumor immune responses result from the infection-induced immunogenic immune-cell death causing epitope spreading [84]. A complex immunotherapeutic strategy is needed for a sustained and robust anti-tumor effect. It is not enough to administer immunostimulatory cytokines, such as IL-2 and IL-15, to elicit innate and adaptive anti-tumor immune responses, primarily involving NK cells and cytotoxic T-cells, respectively. It is now understood that tumors avoid immune-mediated elimination by the secretion of immunoinhibitory molecules, recruitment of immunosuppressive cells, and the upregulation of immunoinhibitory checkpoints such as CTLA-4 and PD-L1 [163]. Studies addressing the questions of how bacteria kill tumor cells are still limited. Additional mechanisms will surely be revealed once data from more studies on this issue become available.

Control of efficacy

One of the major questions affecting bacterial therapy is how to produce a high enough local concentration of a therapeutic. Compared to viruses, bacteria are larger and can be used to deliver multiple gene products. However, bacteria are small relative to mammalian cells (about 1000-fold by volume), meaning that any produced protein must be effective at low concentrations. Regardless of the mode of action, efficacy depends on the local concentration of the therapeutic molecule. A high local titer can be achieved by manipulation of bacterial gene expression, metabolism, or protein secretion. The local tumor concentration could also be enhanced by increasing tumor colonization and proliferation. Better understanding of these mechanisms could greatly improve bacterial therapy.

In comparison, viral therapeutics have limited coding capacity due to small genomes and are limited to delivering only a small number of genes. In the context of delivering an immunomodulatory gene sequence, most viral vectors deliver a single gene (e.g. GM-CSF, IL-12, or IFN-α2b) and conflicting data raise questions as to whether this is sufficient for immune activation robust enough to break peripheral tolerance to tumor antigens. Strategies to address this may include addition of other immunomodulators (e.g. TLR agonists such as imiquimod) to therapeutic regimens, and/or use of viral vectors encoding regulatory RNA sequences implicated in the global regulation of immunosuppression or pathways at the level of the transcriptome.

Control of colonization and delivery

The control of localization and delivery is essential for bacterial therapies. Focusing therapy specifically to malignant tissue can greatly increase efficacy and reduce toxicity to normal tissue. Early clinical trials with bacteria showed that facultative anaerobes colonize human tumors, but at lower densities and with more heterogeneity than in mouse models [150]. Although much is known about the mechanisms that control bacterial tumor colonization, more strategies are needed to harness these mechanisms and make the organisms perform optimally in human tumors.

There are three categories of bacterial therapies based on how they accumulate in tumors: obligate anaerobes (e.g. Clostridium and Bifidobacterium), facultative anaerobes (e.g. Salmonella and Escherichia); and immune-cell-targeting bacteria (e.g. Listeria). The spores of obligate anaerobes germinate exclusively within the hypoxic regions of the cancer [10, 26, 164]. Vegetative anaerobes cannot survive in oxygenate environments of normal organs and can only survive in the anoxic regions of tumors [130, 165]. Facultative anaerobes accumulate in tumors via several interacting mechanisms. After intravenous injection, bacteria flood into tumors following inflammation [166] and become entrapped in the vasculature [167]. Small molecules produced by tumors draw bacteria into the tissue, where they preferentially replicate in the favorable microenvironment [79, 168]. Listeria accumulates in tumors by infecting (MDSC) [113, 169, 170]. These cells are present in blood in large numbers of patients and mice with cancer [171], and are selectively attracted to the TME [172]. Once at the tumor site, Listeria spreads from MDSC into tumor cells by polymerization of actin filaments and production of the pore-forming enzyme listeriolysin O [173]. For all three types of bacteria, immune suppression prevents clearance in the tumor environment but not in normal tissues, which lack immune suppression [174].

Safety

Modern microbial therapies have an excellent safety record. In almost all oncolytic virus trials to date, few adverse events have been attributed to replicating viruses. However, off-target effects and viral mutation/transmission remain ongoing concerns. An open question is the optimal administration of oncolytic viruses. When replicating viruses are administered systemically, non-malignant cells are spared because they retain more robust anti-viral defenses than transformed cells. Infected normal cells either destroy the oncolytic virus or commit apoptotic suicide resulting in a self-limiting infection. As demonstrated by intralesional administration of T-VEC, distant tumor regression suggests that in situ vaccination has primed a systemic immune response (the abscopal effect). It remains unclear if systemic administration can improve response rates without additive or synergistic toxicity. Because human bodies have evolved to fight infections, immunity to microbial agents may develop in response to systemic administration. Major challenges in the development of systemic delivery are serum neutralization and hepatotoxicity. A neutralizing antibody response has been observed in almost every patient treated with viruses, but has not been correlated with a response or lack thereof.

Numerous tumor-targeting bacterial strains have shown a therapeutic benefit in experimental tumor models for decades, and yet clinical development has been slow to follow. Safety is a major concern for therapy with live bacteria, where therapeutic effect is intertwined with toxicity. Preclinical pharmacologic and toxicologic studies with live bacteria have shown satisfactory safety profiles in both healthy and tumor-bearing experimental animals [28, 29]. However, robust intratumoral colonization is required for optimal therapeutic effect. Such an aggressive infection, even confined within the target tumor, will inevitably result in clinical symptoms and laboratory findings that may be considered dose-limiting toxicities (DLTs). This dilemma is perhaps best illustrated in the therapy with oncolytic bacteria. The objective for oncolytic bacterial therapy is essentially to convert a solid tumor into an abscess that forms when an extensive bacterial colonization and subsequent tumor lysis take place [10, 190]. Abscess formation is obviously a desired therapeutic effect and a significant toxicity at the same time. Therefore, whether or not the abscess should be qualified as a DLT is debatable. In this way, it is the very mechanism of tissue-specific pathogenesis that provides the potency to be exploited as a curative. Because of the plasticity of bacteria, the overarching question is whether clever genetic manipulations can be made to utilize endogenous mechanisms and shift the balance toward an effective therapy with minimal safety concerns [98].

Better animal models

Animal models are invaluable to the understanding of neoplastic development and to the evaluation of novel therapeutics. Each model has its benefits and deficiencies which should be carefully considered prior to the study. The implementation of multiple models over the course of the translational development path will aid in recapitulating as many aspects of human biology as possible. The choice of model should balance cost and efficiency with accurate representation of pathology and progression.

Allo- and xenotransplant models provide time-saving experimental protocols and less expensive animals than more complicated models. The use of immunocompromised xenotransplant models can provide information about the efficacy of oncolytic therapies, but are inappropriate for immunotherapies. Immune-deficient mice lack the ability to develop an adaptive memory response against both the microbe and the implanted tumor. Although allo-transplant models are immunocompetent, the speed of tumor growth in transplant models can prevent development of systemic or TME immune states present in human cancer [191].

Genetically engineered mouse models (GEMMs) may more acurately recapitulate essential aspects of human cancer [192]. Autochthonous tumors form in GEMMs from a single cell and grow with a time course similar to human cancer. This slower rate allows more natural structural maturation of the TME [193] and more closely approximate the molecular and histopathological features of human neoplasms [194, 195]. The major limitations of autochthonous tumors in GEMMs are their cost, relatively low mutational load, and reduced clonal heterogeneity. Many GEMMs result in numerous transformed “cancer-seeding” niches, with uneven synchronies but relatively homogeneous genetics, which can make GEMMs hard to cure via microbial therapies.

Patient-derived xenografts (PDX), reconstitute the heterogeneity of the genomic aberrations seen in human cancers. This property overcomes the major limitation of GEMMs, but introduces additional limitations. In this model, primary human tumor tissue is introduced into “humanized” mice that have been reconstituted with a human hematopoietic system [196]. This model can be cost prohibitive and has several limitations. There are inherent deficiencies of the human immune system when it is developed within a mouse, and tumor cells do not completely interact with host stromal and immune compartments in the local microenvironment. This system is also limited by the technical challenge of human leukocyte antigen (HLA) matching implanted tumors with the immune system of the reconstituted mice.

The use of rodents to determine the safety and efficacy of novel agents has inherent limitations because infectious biological agents have co-evolved with mammals and are species-specific pathogens. The inherent resistance of mice to some human pathogens makes preclinical testing of these microbial agents particularly challenging. Alternate models to evaluate the safety and/or efficacy are Syrian hamsters and non-human primates.

Evaluating therapeutic efficacy in veterinary clinical trials is also a useful approach as it permits safety and efficacy testing in large spontaneous tumor models [197–202]. Spontaneous canine and feline cancers represent a naturally occurring heterogeneous model of cancer that resemble human disease in their histopathological and molecular characteristics [203]. Two important aspects of companion animals are that (1) the microbiome of pet dogs shares many features with that of humans [204–207], and (2) the process of immunoediting occurs over the course of months to years, a time frame not achievable in murine models [208]. When considering a large companion animal trial, it is important to consider the frequency and biology of tumors which can sometimes be quite distinct from human disease.

New strategies for microbial-based cancer prevention

Microbial therapies have the potential to be potent cancer prophylactics. To be effective, prophylactic vaccines could (1) target cancers with viral etiology, (2) directly target precancerous lesions, or (3) enhance the immune response against precancerous cells. Some cancers have a viral etiology including liver cancer, cervical, head and neck, anal cancer, Burkitt Lymphoma, nasopharyngeal carcinoma, Merkel Cell Carcinoma, Kaposi Sarcoma, and HTLV-1 associated T-Cell Leukemia/Lymphoma. While the effectiveness of vaccination against cancers with viral etiology has been demonstrated for human papillomavirus (HPV) and hepatitis B (HBV), other vaccines are needed to target additional viruses that are highly associated with cancer, including HCV, MCV, EBV, HHV-8, and HTLV-1. Microbial autologous DNA/RNA vaccines could prevent recurrence by targeting known tumor-associated antigens. Targeting precancerous lesions would prevent the transformation into tumor cells. These vaccines could be prepared to target antigens from patient-derived tumors or human tumor cell lines. Prophylactic vaccines might be useful in cases where there is known family history risk, including known genetic determinants that lead to a high risk of cancer. Microbial-based therapies could also prevent cancer by eliminating cancer-associated microorganisms, or by targeting neovascularization and dormant cancer cells. Microorganisms that target tumor antigens would also be useful for eliminating residual cancer cells after treatment.

GMP manufacture

The GMP manufacture of microbial-based cell therapies is a regulated process that centers on product safety, consistency, and stability. Any investigational new drug (IND) application submitted to the United States Food and Drug Administration (FDA) is required to have a comprehensive description of the chemistry, manufacturing, and control (CMC) information used while generating the product. The FDA has recently published detailed information, in the form of an industry guidance, on the information that must be submitted in an IND pertaining to bacterial- or viral-based biologic products [209].

Critical to the CMC package for any biological product manufactured under GMP conditions is the ability to properly characterize the sterility, purity, identity, potency, and stability of the drug substance and drug product, including its raw materials, and critical intermediates. GMP manufacture of microbial-based cell therapies begins with raw materials and a list of each component used in the manufacturing process. Second to raw materials is the generation of a seed stock, expansion and characterization of the microbial Master Cell Bank (MCB), and any further expansion into a working cell bank (WCB). Key considerations in the manufacture of the seed stock, MCB and WCB include verification of the absence of any lysogenic prophage (for bacteria), genome sequence information characterizing any chromosomal modifications (deletions and additions), phenotypic confirmation of attenuation, microbial purity (clonality), cell viability, and the stability, restriction digest pattern(s), and sequence identity of any episomal constructs (e.g. bacterial expression plasmids). The stability of these features within the MCB and WCB should be characterized over time to gain an understanding of when failure occurs. During early phase production, a master manufacturing record (MMR) is established based on at least one process development experience at the intended manufacturing scale. The MMR is subject to industry standard Quality Assurance (QA) and Quality Control (QC) review procedures and sponsor approval before being accepted for use and employed to officially codify and record manufacturing steps.

Manufacturing processes are typically divided into upstream, downstream, fill/finish, and release testing components. For bacterial cell therapies, the upstream process usually begins with MCB/WCB vial thaw and includes fermentation and formation of the crude cell paste. Downstream processing of bacterial cell therapies terminates after formulation and that process intermediate is defined as the formulated drug substance. A suite of release tests is typically performed on the formulated drug substance prior to advancing to the fill/finish stage. In fill/finish, the drug substance is individually vialed, packaged, labeled as appropriate and is now defined as the drug product. Vialed drug product is subject to further release testing and released for stability testing and use in clinical trials. For non-living bacteria-based products and virotherapies, sterility must also be determined prior to use. With living bacterial cell therapies, demonstrable microbial purity is required.