Bacterial immunotherapy of gastrointestinal tumors

Coley’s historical idea on inducing a strong inflammatory response that leads to tumor reduction provided the basis for developing different forms of immunotherapy. The ideal anti-cancer therapy should selectively eradicate tumor cells, whilst minimizing side effects to normal tissue. Today, it is known that both direct tumoricidal effects and immune activation force antitumor activities. During the immune response, Toll-like receptors (TLRs) sense a diversity of PAMPs to organize the body’s immune defense. This includes (i) stimulation of pro-inflammatory immune cells capable of lysing (infected) target cells in an antigen-independent manner (neutrophils, macrophages) and (ii) inhibition of the tumor-induced immune suppression (tumor-associated macrophages, myeloid-derived suppressor cells and regulatory T cells) that ideally (iii) leads to the development of potent cellular immune responses—often dominated by activated cytotoxic T cells—finally mediating long-term protection against cancer.

Over the past 50 years, several strains of facultative and obligate anaerobic bacteria were applied as oncolytic agents due to their capacity to selectively proliferate in oxygen-starved environments [13‐17]. Based on the observation that necrotic regions exist only within tumors and not in normal tissues, the group of Vogelstein demonstrated that lethal toxin-free Clostridium novyi NT spores are very efficient in eradicating established tumors [14, 18]. They showed that anaerobic bacteria specifically and preferentially target solid tumors, leading to an inflammatory reaction within the tumor that is followed by tumor regression in about 30% of cases [19, 20]. In a preclinical regimen, anaerobic C. novyi NT spores were combined with conventional drugs or radiotherapy. This strategy, referred to as combination bacteriolytic therapy (COBALT), was shown to mediate dramatic and prolonged regression of subcutaneous tumors following a single systemic administration [14]. Similarly, bacteria were found to improve the efficacy of radiotherapy in several mouse models [21]. The authors explained their findings with the fact that efficient tumor cell killing by radiation requires oxygen. Hypoxic cells are more resistant to ionizing radiation than normoxic cells, and hence hypoxic zones in poorly vascularized or necrotic tumors are a major handicap in cancer therapy [22, 23]. By using anaerobic bacteria, these hypoxic regions can be targeted and destroyed, making tumors vulnerable to radiotherapy.

Of particular importance, bacteriolytic therapy was reported to induce immunological memory. C. noyvi infection is associated with inflammation (secretion of neutrophil-directed cytokines) at the tumor periphery leading to development of an effective cellular antitumoral immune response (monocytes and lymphocytes) [14, 17, 18]. Lymphocyte transfer from cured animals into naïve tumor-bearing mice revealed that CD8⁺ cytotoxic T cells are the main cell type involved in this process [20]. However, in a subsequent study performed by our group, using the very same treatment protocol, similar results could not be observed [24]. The bacterial treatment predominantly activated the innate immune system’s arm. NK cells, but not (tumor) antigen-specific T cells, have been activated by bacteriolytic therapy.

Beyond that, at least in our hands, applicability of this approach was imperfect due to troubles in standardization. Regarding toxicity, tumor size and spore dose, we observed that (I) small tumors (< 150 mm³) were completely unaffected; (II) very large tumors (> 450 mm³) responded with substantial necrosis followed by shrinkage but significant animal mortality and (III) an optimal treatment window exists for tumors of approximately 250–300 mm³ [24]. The comparably high mortality rate in large-tumor bearing mice was most likely attributable to the so-called “tumor lysis syndrome” [25]. Similar experiences have already been described by Diaz and colleagues in a large study on evaluating pharmacology and toxicity of C. novyi NT spores [20]. Nevertheless, a clinical trial has been initiated in 2006, but it was terminated due to design problems. Attempts in re-emergence of clostridia-based therapies in the clinical setting are on their way. In August 2011, the BioMed Valley Discoveries Inc. started a pilot study to investigate the safety of C. novyi NT spore administration in patients with treatment-refractory solid tumor malignancies. The primary aim of this study is determination of the optimal time point to initiate antibiotic treatment in these patients. Results of this study will finally help to estimate whether C. novyi NT-based immunotherapies are clinically safe and feasible (Table 1).

Besides C. novyi, genetically modified strains of Salmonella typhimurium have gained interest as potential anticancer agent, either alone or in combination with radiotherapy and chemotherapy [26]. Zhao and coworkers designed a tumor-targeting strain auxotrophic for the amino acids arginine and leucine, termed A1 [3]. In contrast to strictly anaerobic bacteria, S. typhimurium growth is not confined to necrosis and can be found throughout the tumor, including viable regions [27]. There, bacteria invade and replicate intracellular in cancer cells. But growth is not sustained in normal tissue, indicating its potential applicability against small tumors or maybe even micrometastases. If finally necessary, tumor-colonizing salmonella can be readily controlled by early post-infection systemic administration of the antibiotic ciprofloxacin. Tumor lytic effects and accompanying specific antitumor immune response development is preserved [28]. Thus, S. typhimurium-mediated tumor therapy might be applied safely when combined with early antibiotic treatment. To increase the tumor targeting ability and killing efficacy, the A1 strain was further modified by re-isolation from a tumor growing in a nude mouse and termed A1-R [27]. Of note, this strain was able to eradicate metastatic lesions in orthotopic models of breast, prostate and pancreatic cancer, both after local as well as systemic administration [27‐30].

These observations have contributed to the initiation of a clinical phase I trial [31]. Twenty-five cancer patients received bolus infusions of a lipid A-attenuated S. typhimurium (VNP20009). This strain was safely administered to the patients. Focal tumor colonization was observed in some patients receiving high bacterial numbers. None of the patients experienced objective tumor regression, including those with colonized tumors [31]. This is a striking contrast to the results obtained with rodent models. A conceivable explanation is that differences with respect to tumor vasculature, bacterial entry into and growth within tumors, as well as clearance of bacteria exist between rodents and patients. Further studies are required to finally judge if this regimen may still be successful in the clinics.

Irrespective of the presence or absence of intratumoral necrotic lesions, our group showed in a series of experiments that viable as well as lysed and avitalized gram-positive facultative anaerobic bacteria have a direct impact on tumor growth in immunocompetent and in immunocompromised hosts. In an initial work, we explored the antitumoral potential of a bacteriolytic therapy based on S. pyogenes (4). This strain was chosen as a bacterial model because (i) it exerts direct cytolytic effects on eukaryotic cells by secreting several cytotoxic enzymes and pore-forming toxins, (ii) it effectively induces the secretion of proinflammatory cytokines (TNFα, IFNγ) and chemokines (G-CSF), (iii) it mediates high influx of inflammatory cells (e.g., neutrophils, macrophages and NK cells) into the focus of infection and (iv) it initiates the generation of an adaptive, cell-mediated immune response [32, 33] (Fig. 1).

S. pyogenes binds target cells via fibronectin or collagen. Bound fibronectin acts as a bridging molecule towards host cell integrins, which in turn initialize the uptake process that leads to internalization (Fig. 2) [34]. Direct tumor cell contact is thus necessary for S. pyogenes infection. This finally leads to the induction of tumor cell apoptosis. Accordingly, in vivo bacteriolytic therapy in tumor-bearing mice was performed by local injection. A single application of viable bacteria resulted in complete pancreatic carcinoma regression that was accompanied by massive immune activation secondary to infection. As a consequence, immunological memories, as determined by in vitro functional tests and in vivo rechallenge experiments, developed [4]. However, as for potential clinical application, inactivation prior to administration seemed necessary. The streptococcal lysate used in a subsequent study was also shown to mediate substantial growth arrest when injected intratumorally. Again, this antitumoral effect could be attributed to a massive stimulation of immune response mechanisms including a strong systemic elevation of granulocyte numbers and an increase in tumor infiltrating cytotoxic T cells [35]. Taking into consideration that the streptococcal lysate had striking antitumoral activity even when administered alone, it might thus be useful as immunotherapeutic adjuvant in combination with conventional chemotherapy.

Having in mind that the innate immunity is considered to be the sentinel of the first line defense that contributes to the containment and elimination of microbes, we next focused on stimulation of the unspecific immune system, i.e. macrophages, granulocytes, dendritic cells (DC) and NK cells. NK cells in particular can promote tumor regression either directly through the anti-tumor activity of type I interferons or through cell-mediated tumor cell killing. In our study on using avitalized staphylococci, effective tumor growth delay was found to be accompanied by increased numbers of tumor-infiltrating immune cells (NK cells and DC) [36]. The importance of the innate immune system on tumor control could further be corroborated in T cell deficient mice using human colorectal carcinoma (CRC) xenografts. However, despite the observed significant tumor growth control, S. aureus did not provoke complete cure, independent from the treatment regimen [36]. When comparing these findings with previous results using live and lysed bacteria, the adjuvant stimulus of avitalized bacteria was probably strong enough to break the tumor-induced tolerance; however, it is not strong enough to induce long-term immunological responses.

To improve this regimen with respect to specific tumor targeting, we performed experiments using conjugates of bacteria and therapeutic monoclonal antibodies (mAb) for experimental immunotherapy (manuscript in preparation). This strategy is based on the idea that clinically successful therapeutic mAb may have an additional or synergistic impact on tumor growth. In first line treatment, therapeutic mAbs like cetuximab and panitumumab specifically targeting epidermal growth factor receptors (EGFR) are applied in combination with chemotherapies [37‐39]. MAb can have effects on tumor cells by (i) disturbing receptor-mediated signaling, (ii) complement-mediated tumor cell lysis and (iii) cell-mediated tumor cell attack. This may finally lead to a boosted overall immune response. By using immunocompromised mice, we performed clinically relevant systemic injections based on the hypothesis that bacteria coupled to tumor-specific mAb shall permit effective tumor cell targeting in vivo thus concentrating the lytic effects of bacteria on the tumor. Beyond that, potential allergic reactions might be minimized. Although we could not experimentally prove tumor-targeting by bacteria–mAb conjugates, i.e. neither bacteria nor mAbs were detectable in tumors, we observed a substantial delay in tumor growth. This result correlated well with immunological parameters. We found massively increased numbers of circulating monocytes, DCs and NK cells. This provides a basis to refine the concept on bacteria-based combination therapies in the future.

Taken together, these studies contribute to the deeper understanding of how bacteria or their related products act on tumor and immune cells. Hence, bacteriolytic immunotherapy is still a promising alternative strategy for cancer treatment, either alone or in combination with cytostatic drug or antibody therapy. Future investigations will show if these experiments stand clinical trials.

Up to now, Bacillus Calmette–Guérin (BCG) is still the most successful bacterial agent used for superficial bladder cancer treatment [40]. BCG was introduced by Morales in 1976, based on the original development in 1921 as an attenuated strain of Mycobacterium bovis for tuberculosis vaccination. BCG as a live bacterium can exert local (irritative voiding symptoms) as well as systemic (fever, chill, malaise) effects. A functioning immune system is required to prevent excessive reaction, and consequently immunosuppressed patients are not considered for bacterial therapy. Patient’s treatment is conducted on weekly intravesical injections for 6 consecutive weeks. In some cases adjuvant cytokines (IFN) may be added.

The cell wall skeleton of M. bovis is supposed to be the major immune stimulating component. BCG–cell wall skeleton induces IFNγ secretion and stimulates skin Langerhans cells to convert to DCs [41, 42]. Thus, it may act as an ideal adjuvant for immunotherapy. By taking advantage of the favorable immune effects, Mycobacterium tuberculosis strain H37Ra, a heat-killed and dried bacterial preparation, has been widely applied as part of the complete Freund’s adjuvant to attract macrophages and neutrophils to the injection site. Complete Freund’s adjuvant is typically used for initial injections and incomplete Freund’s adjuvant (without H37Ra) for subsequent boosts [43, 44]. Of note, many new candidate immunoadjuvants are tested in clinical trials in order to treat several tumor entities. An overview of actual recruiting trials is given in Table 2.

Bacteria, either used as direct anticancer agent or as a vehicle for cytotoxic drugs, mediate strong pro-inflammatory reactions that have beneficial effects for tumor therapy. On the contrary, there is strong evidence that chronic viral or bacterial inflammation initiates or triggers tumorigenesis [51, 52]. Epidemiological studies showed that a number of chronic infections predispose to various tumor types. In this regard, infection by H. pylori is associated with gastric cancer and mucosal lymphoma, while viral infections are related to cervical and liver cancer [53‐58]. Vaccination against such viruses has proven efficient in preventing cancer development [59, 60].

The underlying mechanism on how bacteria and viruses may instigate malignant transformation is in part attributable to the expression of several toll-like receptor (TLR) agonists, a broad variety of structurally conserved molecules derived from microbes (an overview is given in Fig. 3). On the one hand, TLR stimulation mediates tumor cell apoptosis and accelerates innate immune activation that is often followed by cellular Th1-directed immune responses [61‐64]. On the other hand, TLR stimulation exerts protumoral effects by favoring (i) tumor initiation, development and invasion, (ii) resistance to chemotherapy and (iii) immune tolerance [6, 65‐67]. Indeed, some types of tumor cells exhibit increased proliferation rates upon TLR stimulation [own unpublished data]. These findings fit well with a recent study performed by Cherfils-Vicini and colleagues, showing that TLR7 or TLR8 stimulation of primary lung tumor cells activates NF-kappaB, upregulates Bcl-2 expression, and increases tumor cell survival and chemoresistance [68].

Besides microbial-induced tumorigenesis, non-pathogenic triggers of chronic inflammation, including autoimmune diseases (e.g. inflammatory bowel disease), enhance the risk for malignant transformation (Fig. 4). Accordingly, non-steroidal anti-inflammatory treatments are supposed to decrease tumor incidence [69]. However, a very recent publication describes controversial effect on long-term “chemoprevention” against cancers. The authors showed in a series of large-scale nested case–control studies that long-term use of selective cyclooxygenase 2 inhibitors was associated with a reduced risk of colorectal cancer (CRC), while the risk for breast cancer and haematological malignancies (particularly lymphomas) was increased [70]. These observations underscore the bivalent role of inflammation on cancer, favoring tumor regression in the acute phase while increasing the risk after chronic manifestation.

Two pathways describe the relationship between inflammation and cancer: an intrinsic pathway, driven by genetic alterations that cause inflammation and neoplasia (such as oncogenes), and an extrinsic pathway, driven by inflammatory leukocytes in the context of chronic infectious or persistent inflammatory conditions that augment cancer risk [71]. The latter one is attributable to microbial-induced tumorigenesis. This facilitates a “vicious circle” characterized by (i) local cytokine production (e.g. TNFα, which has controversial roles in cancer, serving as a tumor-promoting or tumor-destructive factor), (ii) chemokine production (CCL7, CCL20, CCL25, CXCL1 and CCL26), (iii) secretion of growth factors (EGF, IGF, TGFα, VEGF), (iv) suppressing adaptive immunity and T cell function and (v) secreting matrix-degrading enzymes (matrix-metalloproteinases, e.g. MMP2 and 9). In solid tumors, immune cells are localized both at the periphery and in the tumor stroma, occasionally invading cancer cell nests. Tumor-associated macrophages (TAM) and T lymphocytes are the most abundant immune population in the tumor-microenvironment, although some eosinophils, mast cells, NK cells and rare DC can be found [72].

Most of the infiltrating leukocyte subsets are effective suppressors of antitumoral Th1-directed immune responses. TAM (also M2 or ‘alternatively’ activated macrophages, phenotype: IL-12^low/IL-10^high) inhibit T-cell activation via secretion of different suppressive mediators, such as IL4, IL13, IL10, TGFβ and indoleamine 2.3-dioxygenase (IDO) [73, 74], as well as MHC class II down regulation. The currently accepted concept implies the M2-polarized myeloid cells promote tumor angiogenesis and invasion [75]. In many but not all human tumors, a high frequency of infiltrating TAM is associated with poor prognosis. But macrophages may act as a double-edged sword in cancer, since these cells can be re-educated to exert antitumor activity [76, 77]. It was demonstrated that IL12 injection induced a pro-immunogenic potential of TAM while reducing tumor-supportive activities [78].

In the context of an inflammatory response, myeloid cells are the primary recruited effectors. In tumors, these cells are represented by myeloid-derived suppressor cells (MDSCs). Murine MDSCs are referred to as CD11b⁺Gr1⁺, which are the predominant population of tumor-associated myeloid cells. Similar to TAMs, increased numbers appear in the blood during tumor growth in mice (Fig. 5), with some being recruited to the tumor site to promote angiogenesis [73, 79]. Of note, myeloid cells infiltrating tumors may induce tumor cell resistance to CTLs by modifying the peptide–MHC complexes on tumor cells via reactive oxygen species [79]. These modifications reduce the capacity of the tumor cells’ MHC to bind antigenic peptides and subsequent recognition by CTLs. This provides a novel concept regarding tumor escape associated with inflammation and might have consequences for therapeutic approaches focusing on pharmacological reactive oxygen species inhibition [79].

Finally, regulatory T cells (phenotype: CD4⁺CD25⁺FOXP3⁺CD127^low) which have been identified long time ago contribute to the general tumor-specific T cell tolerance. Tregs are activated in an antigen-specific manner but are believed to suppress T cells in an antigen-non-specific manner [71]. While Treg accumulation at high density in tumors is generally related to a poor outcome, in patients suffering from CRC, high Treg infiltration is associated with a favorable clinical prognosis [80]. To explain this irony, the dense microbiological flora present in the gut requires a T-cell-mediated inflammatory anti-microbial response that, among others, involves Th17 cells [81]. This Th17-cell-dependent proinflammatory and tumor-enhancing response can be attenuated by Tregs, thus reconstituting the balance between pro-inflammatory and anti-inflammatory effects in the gut that may contribute to the favorable role in CRC prognosis. The link between a high density of FOXP3-positive Tregs in CRC may lead to a paradigm shift and thus help to decide when immunotherapy based on the integration of bacteria or microbial structures is feasible and safe.