on-line at: www.actabp.pl Salmonella and cancer: from pathogens to therapeutics*

Bacterial cancer therapy is a concept more than 100 years old - yet, all things considered, it is still in early development. While the use of many passive therapeutics is hindered by the complexity of tumor biology, bacteria offer unique features that can overcome these limitations. Microbial metabolism, motility and sensitivity can lead to site-specific treatment, highly focused on the tumor and safe to other tissues. Activation of tumor-specific immunity is another important mechanism of such therapies. Several bacterial strains have been evaluated as cancer therapeutics so far, Salmonella Typhimurium being one of the most promising. S. Typhimurium and its derivatives have been used both as direct tumoricidal agents and as cancer vaccine vectors. VNP20009, an attenuated mutant of S. Typhimurium, shows significant native toxicity against murine tumors and was studied in a first-in-man phase I clinical trial for toxicity and anticancer activity. While proved to be safe in cancer patients, insufficient tumor colonization of VNP20009 was identified as a major limitation for further clinical development. Antibody-fragment-based targeting of cancer cells is one of the few approaches proposed to overcome this drawback.


INTRoducTIoN
It was at the end of the 19 th century that bacteria were for the first time intentionally applied for cancer treatment.William Coley, a bone sarcoma surgeon at the Memorial Hospital in New York, having conducted a thorough literature search, found 47 well-documented cases of beneficial influence of serious bacterial infections in tumor patients (McCarthy, 2006).On the basis of the then-available data he concocted a mixture of killed Streptococcus pyogenes and Serratia marcescens bacteria, which became known as Coley's toxin.Over the next forty years Coley applied his toxin to almost 1000 patients who suffered from various types of inoperable cancers.Striving to improve the effectiveness of the therapy, he used various doses of his toxin, different regimens and durations of treatment, as well as different routes of application and therefore it is difficult to fully comprehend his enormous work.In many cases several-month treatment resulted in complete tumor regression.The fiveyear survival period for patients suffering from inoperable carcinomas ranged from 34 to 73%, and for those suffering from inoperable sarcomas was in the range between 13 and 79%, varying with the tumor subtype (Green & Hoption Cann, 2007).However, the results of Coley's treatment have been non-reproducible, uncertain, and unpredictable and therefore his therapy often met with strong criticism from the medical community.At the beginning of the 20 th century it was gradually displaced by newly developing radiotherapy, which resulted in fast tumor destruction and pain relief, although not necessarily in complete tumor eradication, especially at the stage of advanced, metastatic disease.
With the progress of immunology it became clear that the mechanism of action of Coley's toxin involves activation of the immune system and a multilevel modulation of immune response.This understanding restored interest in possible therapeutic applicability of Coley's approach.Richardson and coworkers (1999) compared the effectiveness of Coley's toxin with contemporary cancer therapies based on published results concerning patients treated with Coley's toxin and matched controls from National Cancer Institute's Surveillance Epidemiology End Result database (Richardson et al., 1999).They found higher rates of ten-year survival of Coley's patients compared to patients subjected to modern treatment in kidney cancer (33 vs. 23%), ovarian cancer (55 vs. 29%), and sarcoma (50 vs. 38%), which gives food for thought.The attempts to reevaluate Coley's concept are undertaken anew.In 2012 the new phase 1 clinical trial investigating the safety and the dosage of biochemically well-defined and good manufacturing practice (GMP)compliant Coley's toxin, presently known as MBV (mixed bacterial vaccine) was started in the Ludwig Institute for Cancer Research.It involves patients suffering from cancers expressing NY-ESO-1 antigen, including metastatic melanoma, head and neck carcinoma, sarcoma and prostate cancer (Karbach et al., 2012).The results of the trial have not been published yet.
In line with Coley's concept of using bacteria in cancer therapies, other bacterial species were evaluated for a possible anti-tumor effect.Early experiments on "hemorrhagic allergy" in 1910s and 1920s showed that animals injected with bacterial filtrates developed hemorrhagic necrosis upon re-challenge with bacteria (so called Sanarelli-Shwartzman phenomenon).In particular, this condition was caused by Bacillus typhosus, which used to be a synonym for Salmonella Typhi at that time (Shwartzman, 1928).Interestingly, this shock syndrome induced a therapeutic effect in tumor-bearing animals (Shwartzman, 1935).However, the studies were discontinued due to very high treatment-related mortality; nowadays the adverse effects can be explained by strong stimulation of proinflammatory cytokines including tumor necrosis factor (TNF) by endotoxin and other bacterial products.Bacillus Calmette-Guerin (BCG), attenuated bovine tuberculosis bacteria (Mycobacterium bovis), has been used for decades as a vaccine protecting against tuberculosis.However, it also appeared to be one of the most successful cancer immunotherapeutics.BCG, in the form of repeated intravesical instillations, has already been in use for over 30 years as a standard method to prevent cancer recurrence after endoscopic surgery of intermediate-and high-risk non-muscle invasive bladder cancer (Kawai et al., 2013).It is also effective against inoperable bladder carcinoma in situ resulting in a 70-75% complete response rate (Alexandroff et al., 1999).Unfortunately, in other tumor types BCG application has not proved to be more effective than conventional therapies (Alexandroff et al., 1999).

BAcTeRIAl cANceR TheRAPIeS cAN AddReSS The Key ISSueS IN cANceR TReATMeNT
Starting from 1946, chemotherapy gradually became the principal therapeutic strategy in cancer and bacterial therapies were largely forgotten.The success of smallmolecule drugs in cancer treatment was remarkable, but turned out to be incomplete.Chemotherapy of many tumors suffers from limited efficacy towards cancer cells and damaging action towards normal cells.Both phenomena have a significant impact on therapeutic outcomes, being responsible for incomplete tumor killing and adverse effects in other tissues.The two factors responsible for those clinical drawbacks are: low specificity towards cancer tissue and insufficient penetration of the tumor by chemotherapeutics.Cancer cells form a complex and heterogeneous system with areas of high metabolism, rich in nutrients and oxygen, as well as distal regions of poor perfusion, quiescence and necrosis (Saunders et al., 2012).
The use of bacteria as anticancer agents might have multiple advantages over other therapeutic approaches.Bacterial therapies can benefit from microbial metabolism, motility and sensitivity to address a number of issues related to currently used treatment modalities.One of the most important issues, relevant to virtually all chemotherapeutic and biological treatments, is limited accessibility of the tumor tissue to passively-distributed therapeutics.Both in the case of small molecule drugs, as well as larger molecules -cytokines, antibodies or even viruses -the therapeutic agent diffuses from the bloodstream into the periphery, with no transport system that could cross biological barriers, act against haemodynamic gradients or accumulate preferentially in the tumor tissue.In fact systemic delivery of passive therapeutics produces relatively large drug concentrations in the bloodstream and relatively low drug concentrations in the tumor, resulting in limited efficacy and increased extratumoral toxicity.Bacteria offer unique mechanisms that can facilitate site-specific treatment, highly focused on the tumor and safe to other tissues.This can be possible due to several features of bacteria-based therapeutics:

environmental sensing
Tumor tissue has a complex and heterogenic metabolism that makes it particularly resistant to systemic treatment.The natural ability of bacteria to receive signals via chemoreceptors can be used to effectively target this unique microenvironment.Oxygen concentration is one of the most important signals for anaerobic bacteria and is of particular interest in anticancer therapy since hypoxia is a common feature of tumors.Moreover, auxotrophic bacterial strains that rely on the uptake of certain metabolites can recognize the tumor microenvironment as a source of nutrients.This phenomenon can facilitate specific accumulation of bacteria in the tumor.One of the most effective examples is the use of Clostridium spores that can only germinate in oxygen-free tumor regions (Dang et al., 2001).

Motility
The bacterial microorganism is not only capable of detecting chemoattractants, but can also actively follow chemical gradients.This contrasts with passive therapeutics that simply diffuse into tissues from the circulation.Bacteria are able to penetrate deep into the tumor tissue and perform specific actions, e.g.express proteins or transfer genes, to tumor cells localized remotely from the vasculature.This feature can also allow bacteria to cross physiological barriers and accumulate in cellular regions that are either distant and inaccessible for passive therapeutics or quiescent and unresponsive to chemotherapy.For example, motile strains of Salmonella were shown to effectively penetrate tumor tissue in vitro (Toley & Forbes, 2012).However, the role of bacterial motility in in vivo tumor localization is unclear (Stritzker et al., 2010).

Active delivery
Unlike chemical or biological molecules, microorganisms are metabolically active and are able to perform specific metabolic tasks at the tumor site.These include the production of cytotoxic agents (e.g.bacterial toxin), expression of immunomodulatory molecules (e.g.cytokine) or enzymatic conversion of a prodrug into an active therapeutic.Strains derived from intracellular pathogens can infect tumor cells and deliver specific proteins or genes to the tumor tissue (St Jean et al., 2008).Nevertheless, the intratumoral action is not always necessary as bacteria can also express tumor-related antigens to stimulate systemic anticancer immune responses.A Listeria monocytogenes-based cancer treatment that has recently entered clinical development is an example of this approach, where the bacterium delivers tumor antigens directly to the antigen presenting cells (Singh & Wallecha, 2011).

controlled propagation
Preferential tumor growth exhibited by many bacterial species is not exclusive to bacteria; experimental oncolytic viral therapies are based on a similar principle.However, once administered to the patient viral vectors are beyond external control.In contrast, bacterial therapeutics are susceptible to antibiotic treatment and there-U n c o r r e c t e d P a p e r i n P r e s s Salmonella and cancer: from pathogens to therapeutics fore fully manageable in the clinical setting -therapy can be stopped at the onset of adverse effects or when the bacteria are no longer needed.In fact the use of live biotherapeutics that contain antibiotic resistance genes in clinical trials is not recommended by the regulatory agencies1 .

Immunostimulation
Bacterial vectors augment the anti-tumor immune response not only because of their cargo but also due to their own potent immunostimulatory activity.A growing tumor creates an immunosuppressive environment and establishes immune escape mechanisms that limit the maturation of dendritic cells (DCs) as well as the priming and migration of specific T cells into the tumor.Bacteria provide a strong danger signal to the immune system.The specific conserved bacterial structures such as components of the cell wall or unmethylated CpG ssequences ites in bacterial DNA constitute so-  (Dai et al., 2009).
called pathogen-associated molecular patterns (PAMPs) that are recognized by Toll-like receptors expressed by innate immune cells.PAMPs activate innate-and initiate adaptive immune responses.The induction of TNF, IFNg and IL-12 results in the recruitment and activation of DCs which upon migration to the lymph nodes may efficiently present tumor antigens to T cells (Chorobik & Marcinkiewicz, 2011).It has been shown that indeed microorganisms colonizing tumors and promoting an inflammatory reaction in the tumor microenvironment potentiate the anti-tumor host response (Avogadri et al., 2005).
The unique features of bacterial therapeutics create the opportunity for novel anticancer strategies, that combine tumor-related molecular gradients, natural bacterial features and genetic engineering.Bacteria meet all the requirements for an ideal tumor-targeting agent and might become a novel tool in the anticancer toolbox (Forbes, 2010).A summary of the bacterial strains studied in cancer treatment is shown in Table 1.

Salmonella hAS A NuMBeR of feATuReS fAVoRABle foR cANceR TheRAPy
Salmonella belongs to the Enterobactericae family, a group of Gram-negative, facultatively anaerobic and facultatively intracellular pathogenic bacteria.Currently, based on genome sequence similarity, the genus Salmonella is categorized into two species S. bongori and S. enterica which in turn is divided into six subspecies including S. enterica subsp.enterica (Tindall et al., 2005).According to the White-Kauffman-Le Minor scheme, the subspecies are classified into more than 2 500 serovars by serotyping: O-antigens (polysaccharide domain of the cell surface LPS); H1, H2 antigens (flagellin proteins) and the Vi antigen (Guibourdenche et al., 2010).New classification methods including genome-based techniques are currently under development.
In humans, Salmonella enterica subsp.enterica serovars Typhimurium and Typhi are causative agents of gastroenteritis and typhoid fever, respectively.There are more than 27 million cases of typhoid fever worldwide each year with a mortality rate of 0.8% resulting from intestinal perforation and peritonitis or severe toxic encephalopathy connected with myocarditis and hemodynamic shock (Parry et al., 2002).Infections with S. enterica serovar Typhimurium result in an estimated 94 million cases of gastroenteritis worldwide and 0.16% mortality (Feasey et al., 2012).S. Typhi is an exclusively human pathogen, while hosts of S. Typhimurium include rodents, poultry and cattle.
Salmonella Typhimurium and Salmonella Typhi are closely related serotypes of S. enterica species which differ in host adaptation and the outcome of infection.Depending on the serotype and host, Salmonella colonizes solely the intestinal epithelium leading to gastroenteritis or spreads beyond the gut mainly to liver and spleen causing typhoid fever.S. Typhimurium infection in humans is restricted to the digestive tract, with the exception of infants, elderly or immunocompromised individuals in whom it can spread, while in mice S. Typhimurium causes enteric fever.On the contrary, S. Typhi causes typhoid fever in humans but is not pathogenic to animals.In general, serotypes that lack host specificity, such as S. Typhimurium, are more frequently associated with disease in young rather than in adult animals, suggesting their non-optimal adaptation to mature immune system (Baumler et al., 1998).
About 90% of the genes in S. Typhi and S. Typhimurium serovars are identical (McClelland et al., 2001), but among about 4 000 genes of S. Typhi, more than 200 are functionally disrupted or inactive, while most of their homologs are still fully functional in S. Typhimurium.Genes that differ include virulence factors that determine the pathogenic potential, which can in part explain the restricted host range of S. Typhi (McClelland et al., 2001).The majority of virulence factors is encoded by genes grouped in a few clusters in the genome, termed Salmonella Pathogenicity Islands (SPIs).S. Typhimurium and S. Typhi genomes share 11 SPIs, one is specific for S. Typhimurium (SPI14) and four are specific for S. Typhi (SPI7, 15, 17 and 18) (Kolyva, Waxin, and Popoff, 1992).
Salmonella sp. has an ability to multiply inside phagocytic and nonphagocytic cells including macrophages, dendritic cells (DCs), neutrophils, M cells and epithelial cells (Malik-Kale et al., 2011).The ability of Salmonella to invade and survive within a host cell is dependent on two Type III Secretion Systems (T3SS), the multiprotein complexes with a needle-like structure present on the bacteria cell wall.Proteins involved in the assembly of the two major T3SSs of Salmonella are encoded by SPI1 and SPI2.T3SS1, encoded by SPI1, is required for efficient invasion of nonphagocytic cells.In contrast, the expression of SPI2-encoded T3SS2 is induced following the internalization of Salmonella into host cells and is required for post-invasion processes (Velge et al., 2012).During the invasion, some SPI1 encoded proteins such as InvG, InvJ, PrgH, PrgI, PrgK and SpaO are responsible for the assembly of the needle complex, whereas others, including SipB, SipC and SipD, translocate the effector proteins through this needle.Regulation of gene expression in response to the surrounding microenvironment depends on several two-component systems, such as PhoQ/PhoP whose expression is induced by Mg 2+ starvation and low pH (Lucas et al., 2000).Another system, OmpR-EnvZ, responds to changes in osmolarity and regulates invasion (Bajaj et al., 1995) as well as intracellular survival.Upon internalization, Salmonella modifies the phagosome into a Salmonella-Containing Vacuole (SCV), which is characterized by the presence of some lysosomal membrane proteins, low pH, and transient interactions with the endocytic pathways.At early stages of maturation, SCV recruits and quickly looses early endocytic markers, such as the early endosomal antigen 1 (EEA-1) or transferrin receptor (TfR) (Steele-Mortimer et al., 1999).At later points of maturation it acquires several late endosomal markers, including LAMP1 (Lysosomal-Associated Membrane Protein 1) (Steele-Mortimer, 2008).Bacterial replication is accompanied by the formation of dynamic membrane tubules termed Salmonella-Induced Filaments, which extend from SCV throughout the cell (Schroeder et al., 2011).The maturation of SCV is controlled by SPI2-encoded effectors which allow bacteria to avoid phagosome-lysosome fusion and degradation and protect them against reactive oxygen and nitrogen species (Chakravortty et al., 2002;Janssen et al., 2003).
After the ingestion of S. enterica, the bacteria use different routes to cross the intestinal barrier.The main route leads through the receptor-mediated endocytosis by microfold cells (M-cells) in Peyer's patches (Jepson & Clark, 2001), independently of SPI1 and SPI2 (Martinez-Argudo & Jepson, 2008).Then the bacteria are taken up by the underlying macrophages.Enterocytes of the intestinal epithelium, except for the M cells, engulf Salmonella by macropinocytosis in a SPI1-dependent manner.The U n c o r r e c t e d P a p e r i n P r e s s bacteria can be also engulfed by DCs (Swart & Hensel, 2012).In epithelial cells S. Typhimurium can reside and replicate in SCV, as well as in the cytosol (Malik-Kale et al., 2012); however, these two intracellular populations of bacteria are transcriptionally distinct: the intravacuolar bacteria are SPI2-induced, while the cytosolic bacteria are SPI1-induced and flagellated (Knodler et al., 2010).Salmonella induces caspase-1 and -2 mediated apoptosis of infected macrophages and epithelial cells (Jesenberger et al., 2000;Kim et al., 1998;Monack et al., 1996).Apoptosis depends mainly upon the SPI1 effector protein, SipB, delivered to the host cell by SPI1-encoded T3SS (Jesenberger et al., 2000).Bacteria, engulfed by macrophages and DCs spread from Peyer's patches to mesenteric lymph nodes, spleen and liver.The ability of Salmonella to elicit systemic disease is serovar-dependent and correlates with its capability to survive and replicate inside the host cell and to avoid the host adaptive immune response (reviewed by Swart & Hensel, 2012) TheRAPeuTIc Salmonella STRAINS ARe deRIVed fRoM S. TyPhIMuRIuM oR S. TyPhI S. Typhimurium infection in mice remains the dominant animal model of typhoid fever because it leads to comparable systemic disease with dissemination of bacteria to the lymphatic system and peripheral organs.Hence the prevalence of S. Typhimurium-based tumor therapeutic vectors studied in murine models of cancer.The choice of S. Typhimurium over S. Typhi for preclinical research on cancer therapy could be arguable but is legitimated by the availability of suitable animal models.Some improvement was made by the development of a transgenic mouse model of S. Typhi infection (Song et al., 2010), but it has not been applied to Salmonellabased tumor therapy studies yet.Nevertheless, a superior induction of immune response to heterologous antigen delivered by an orally administered vaccine strain of S. Typhimurium over S. Typhi was shown in a clinical trial (Angelakopoulos & Hohmann, 2000).Both serotypes are amenable to genetic modifications of virulence and in terms of safety are equally eligible as live attenuated therapeutic strains.There has already been some success in the clinical use of attenuated Salmonella strains.S. Typhimurium VNP20009 was shown to be safe when administered to cancer patients (Toso et al., 2002) and S. Typhi Ty21a, a live attenuated oral vaccine against typhoid fever, has already been applied for more than 30 years to adults and children above 6 years of age who are at risk of S. Typhi exposure.A phase I clinical trial evaluating the safety, tolerability, and effects of the S. Typhi Ty21a strain used as a DNA delivery vehicle in cancer patients has been announced (Niethammer et al., 2012) but its results are not yet available.
Attenuation of virulence is crucial for the development of new Salmonella-based vector strains in order to elicit an appropriate profile of the immune response.Up to date about 50 genes of Salmonella spp.have been proven to be feasible for inactivation in order to obtain an attenuated derivative with modified virulence or metabolic functions.Gene inactivation is achieved by laboratory selection of the desired phenotype, for example by passaging bacteria through selective conditions and screening for survivors, or is the result of site-directedor chemical mutagenesis.Inactivation of genes coding for proteins involved in metabolic pathways generates auxotrophic strains, i.e. dependent on external sources of nutrients, for example aromatic amino acids (aro mu-tants) or purines (pur mutants).Direct attenuation of virulence involves inactivation of genes encoding proteins interacting with the infected organism or factors regulating their expression.The latter include: (i) phoP and/ or phoQ, which regulate expression of many genes, e.g.those contributing to the resistance against antimicrobial peptides and genes located in pathogenicity islands, (ii) cya and crp genes coding for global regulatory factors involved in expression of many proteins of cellular catabolism, (iii) htrF gene, which enables survival under stress conditions (Garmory et al., 2002;Raupach & Kaufmann, 2001;Raupach et al., 2003).The attenuating mutations introduced into Salmonella experimental strains are summarized in Table 2.
An interesting example comes from a sophisticated work of Robert Hoffman's group: A1-R is an auxotrophic Salmonella strain developed for increased tumor targeting and limited toxicity.First, the researchers obtained an A1 auxotrophic strain dependent on an external source of leucine and arginine by nitrosoguanidine (NTG) mutagenesis of S. Typhimurium 14028-GFP.To further enhance tumor targeting of S. Typhimurium A1, bacteria were injected i.v.into nude mice bearing HT-29 human colon adenocarcinoma.GFP-expressing bacteria isolated from the excised infected tumors, were termed A1-R and had an increased ability to adhere to cancer cells; the number of A1-R bacteria attached to HT-29 cells in vitro was approximately six times higher than that of the parental A1 strain (Zhao et al., 2006).
Applicability of A1-R in metastatic disease has also been tested.Hayashi et al. (2009b) developed a lymph node metastasis model of human pancreatic cancer (XPA-1) by injecting XPA-1 cells into the inguinal lymph node of nude mice which resulted in tumor growth in the axillary lymph node.Five of six mice had their lymph node metastases eradicated within 7-21 days after intravenous treatment with A1-R in contrast to growing metastases in the untreated control group (Hayashi et al., 2009b).
A1-R has also been tested in immunocompetent mice bearing murine Lewis lung carcinoma (Tome et al., 2013).The authors proposed to apply a priming dose of 1×10 6 cfu followed, 4 hours later, by a therapeutic dose of 1×10 7 cfu of A1-R.The priming dose resulted in an elevated serum level of TNF and tumor vessel destruction, which could facilitate the invasion of tumor by bacteria.
As it can be expected, the mode of attenuation affects the quality of the anti-Salmonella immune response.While cytokines of the Th1 type (IFNγ, IL-12, IL-18) are crucial for protection against Salmonella in mice and humans, different mutant Salmonella strains require alternative cytokines for the control of infection.Studies with knockout mice showed that IFNγ and TNF were essential for the early control of infection with both wild type and aroA mutant strains.In contrast, TNF was not required for the clearance of the aroA mutant (Raupach & Kaufmann, 2001).Attenuated Salmonella strains preferentially colonize solid tumors and inhibit their growth in animal models (Bermudes et al., 2002;Pawelek et al., 1997;Zhao et al., 2005).In some therapeutic approaches attenuated strains were used without any further modifications to exert tumor-directed cytotoxic effects and induce proper antitumor immune response.Apart from the A1-R strain described above also the application of the S. Typhimurium Χ9241 strain brought successful results.Bacteria injected intratumorally to CT26 tumors or CT26 tumors expressing human tumor antigen, Her-2/neu, significantly inhibited tumor growth.Surprisingly, they did not potentiate tumor-antigen-specific cellular immunity.However, they induced an important functional shift in the phenotype of tumor-infiltrating CD11b + Gr-1 + subpopulation of leukocytes (myeloid derived suppressor cells, MDSC) towards immunogenic TNF-secreting neutrophils.Moreover, the therapy reduced the percentage of regulatory T cells which may promote tumor development (Hong et al., 2013).
Thanks to the intracellular lifestyle and immunomodulatory properties attenuated Salmonella strains are also used as vectors for the delivery of therapeutic molecules.The cargo is a genetic material which codes for proteins including tumor antigens (described in the next chapter), cytokines, apoptosis-inducing factors, prodrug-converting enzymes, or short hairpin RNAs (shRNAs) able to silence expression of a protein of choice.Numerous approaches were developed to enhance and navigate the immunomodulatory properties of Salmonella through genetic modifications that ensure the delivery of proapoptotic molecules or cytokines straight to the tumor, which limits their potential undesired systemic side effects.The examples are listed in Table 3.
The group of John C. Reed demonstrated a significant increase in tumor growth inhibition by a Salmonella strain expressing IL-18 in immunocompetent mice (Loeffler et al., 2008).The modified and parental strains showed comparable toxicity limited to the spleen and liver.IL-18 stimulates T cell and NK cell to proliferation, cytotoxicity and cytokine production.Administration of IL-18-secreting Salmonella did not improve tumor infiltration by CD8 + T lymphocytes but led to increased infiltration by granulocytes, CD4 + T cells and DX5 + NK cells, when compared to Salmonella control strain (Loeffler et al., 2008).
In order to enhance the proapoptotic activity of S. Typhimurium towards infected cancer cells several research groups took an advantage of apoptosis-inducing factors and equipped bacteria with appropriate expression vectors.TRAIL (TNF-related apoptosis-inducing ligand) has a capacity to selectively induce apoptosis in a wide variety of cancer cells but hardly in normal cells which makes it a promising cancer therapeutic (Hylander et al., 2005;Shanker et al., 2008;Walczak et al., 1999;Yagita et al., 2004).Similarly to TNF, TRAIL exerts its proapoptotic effect through the death receptor-dependent pathway which activates the caspase cascade (LeBlanc & Ashkenazi, 2003).Ganai et al. (2009) constructed an expression vector placing TRAIL under the promoter of recA gene involved in the prokaryotic SOS response to DNA damage (Anderson & Kowalczykowski, 1998).It created a radiation-inducible system, where expression of TRAIL was turned on by genotoxic damage evoked by radiation.Such a system enables temporal and spatial control of the gene expression, additionally increasing the selectivity of the therapy towards cancer cells subjected to radiotherapy.Intravenous administration of S. Typhimurium VNP20009 equipped with a PrecA-TRAIL construct (VNP/pRE-TR) into Balb/c mice bearing subcutaneous 4T1 mammary carcinoma followed by 2 Gy whole body γ-irradiation 2 days later resulted in a significant decrease in tumor growth.The combined treatment increased the
A similar approach was exploited by Chen et al. (Chen et al., 2012) who placed TRAIL cDNA under the control of nirB promoter induced by hypoxic conditions (Chatfield et al., 1992).The S. Typhimurium VNP20009 strain carrying TRAIL expression vector was intraperitoneally U n c o r r e c t e d P a p e r i n P r e s s administered to B16F10 melanoma-bearing C57BL/6 mice.Significant inhibition of melanoma tumor growth and extended survival time were observed.The TUNEL assay showed that the therapeutic effect of modified bacteria was associated with a significant increase in apoptosis of melanoma cells compared to mice receiving the control strain.Importantly, the immunohistochemical study revealed that indeed TRAIL was preferentially expressed in the hypoxic, necrotic area of the tumor, and not in the oxygenated liver or spleen, which may explain limited toxicity of TRAIL in normal tissues.
It has been shown that Smac (second mitochondriaderived activator of caspases) is involved in TRAIL-induced apoptosis and may affect the efficiency of TRAILbased therapies (Deng et al., 2002;Zhang et al., 2001).Also in vivo studies showed that complete regression of established glioma in a mouse model was achieved only when TRAIL therapy was associated with Smac administration (Fulda et al., 2002;Pei et al., 2004).Fu et al. (2008) took an advantage of this observation and engineered a modified S. Typhimurium SL3261 strain carrying a dual-gene vector coding for both Smac and TRAIL.The cDNAs were placed under the promoter of human telomerase reverse transcriptase, highly active in majority of cancers and generally inactive in normal differentiated cells (Kim et al., 1998).LL/2 Lewis lung carcinoma, B16F10 melanoma and 4T1 mammary carcinoma cells infected with the modified bacteria showed high expression levels of both proteins accompanied by a high rate of apoptosis.Moreover, oral administration of the modified strain significantly suppressed tumor growth in all tested mice models (LL/2, B16F10, 4T1), without any observable side-effects.The volumes of the tumors were lowered approximately by 65-75% compared to groups treated with a control strain and by 90% compared to PBS-treated animals.Cao et al. (2010) applied combination of TRAIL with another apoptosis-inducing factor -apoptin (VP3), a protein of the chicken anemia virus proved to promote tumor cell-specific apoptosis in a p53-independent manner (Zhuang et al., 1995).The study showed that intratumoral administration of S. Typhimurium SL7207 carrying apoptin and TRAIL coding sequences under the control of eukaryotic cytomegalovirus immediate early promoter to human gastric tumor xenografts in nude mice resulted in increased effectiveness of bacteria-mediated tumor growth suppression, with complete tumor regression in some animals.TUNEL assays in tissue sections showed a higher apoptotic rate in mice treated with S. Typhimurium SL7207 carrying apoptin and TRAIL genes compared to a control strain (Cao et al., 2010).

Salmonella cAN Be AlSo uSed AS A TuMoR VAccINe VecToR
The vast majority of tumors express proteins or other antigens that are absent (or present only in very low quantities) in healthy adult tissues.These tumor-associated antigens (TAAs) are potentially immunogenic and tumor development is usually accompanied by specific, although often ineffective, anti-TAA immune response.TAA vaccines used for cancer therapy often fail, probably due to inadequate antigen presentation and insufficient activation of innate immunity.The application of Salmonella as a vector for TAAs should result in overcoming both impediments.The first attempts to deliver TAA via Salmonella were undertaken in late 1990s.From that time numerous studies utilizing natural (mPSCA, mAFP, survivin, endoglin) or artificial (β-galactosidase) tumor antigens have proved that placing a TAA-coding transgene under strong cytomegalovirus promoter in a plasmid carried by Salmonella allows for TAA expression in the cytoplasm of infected cells or dendritic cells which engulfed the infected, apoptotic cells (Paglia et al., 1998;Yrlid & Wick, 2000); TAA expression elicits efficient cell-mediated-or both cell-mediated and humoral immune responses (Ahmad et al., 2011;Chou et al., 2006;Fest et al., 2009;Huebener et al., 2008;Jarosz et al., 2013;Paglia et al., 1998).However, this is not always the case.The group of Dai-Ming Fan transformed Salmonella with a plasmid, which expressed a fusion protein consisting of a mimotope of the gastric tumor antigen, MG2 with either PADRE T helper epitope (Guo et al., 2003) or HBcAg, a strong Hepatitis B virus T-cell-dependent and T-cell-independent antigen (Meng et al., 2005).Although oral administration of either Salmonella strain resulted in a partial inhibition of the growth of the subsequently injected Ehrlich ascites carcinoma, only the humoral but not the cell-mediated response against MG2 could be detected (Guo et al., 2003;Meng et al., 2005).
It is believed that intracellular location of Salmonella within phagosomes may limit the ability to generate a MHC class I-restricted immune response towards transgene-encoded proteins.However, Salmonella expresses a multi-protein complex, T3SS, to deliver some bacterial effector proteins into the host cell in order to modulate its functions and create a microenvironment favorable for bacterial survival and proliferation (Galan and Collmer, 1999).A special N-terminal signal sequence directs bacterial proteins for export through T3SS to the host cell cytoplasm.This very mechanism was employed by several research groups to deliver Salmonella plasmidencoded tumor antigens to the host cell cytoplasm and thus make them available for the MHC class I presentation.Instead of sequences encoding TAAs alone, the constructs coding for chimeric proteins consisting of a tumor antigen preceded by an N-terminal fragment of a bacterial protein subjected to T3SS export (containing secretion and translocation signals) were placed in the Salmonella plasmid.These included peptides derived from the SopE, SseF and YopE proteins (Jellbauer et al., 2012;Manuel et al., 2011;Nishikawa et al., 2006;Zhu et al., 2010).The chimeric transgenes were usually placed under bacterial promoters activated after invasion of the host cell.
This approach was applied by Nishikawa et al. in an attempt to generate a therapeutic vaccine against cancers overexpressing the NY-ESO-1 antigen (Nishikawa et al., 2006).This germ cell antigen seems to be a proper target for immunotherapy, since it is expressed by a broad variety of tumors but not by normal somatic cells.Following oral administration of the Salmonella-based vaccine that delivered NY-ESO-1 and assured its cytoplasmic localization, CD8 + T cell-dependent regression of established NY-ESO-1-expressing tumors was observed.What is more, the vaccine was able to initiate a process known as epitope spreading and was therefore effective even towards tumors lacking the NY-ESO-1 antigen.Intratumoral injection of this Salmonella strain delivering the antigen and engaging pre-existing NY-ESO-1-specific CD8 + T cells resulted in tumor regression and activation of subsets of T cells recognizing at least two different tumor antigens that were not delivered by Salmonella (Nishikawa et al., 2006).
A more complex transgene was introduced into a Salmonella plasmid by Zhu et al., who worked on an efficient anti-melanoma vaccine (Zhu et al., 2010).It en-U n c o r r e c t e d P a p e r i n P r e s s Salmonella and cancer: from pathogens to therapeutics coded a fusion protein consisting of the secretion and translocation signals of SopE and a fragment of the melanoma-specific TRP2 (tyrosinase-related protein 2) antigen containing three immunodominant epitopes.The presence of Hsp70, often referred to as immunochaperone, facilitates the proper presentation of antigenic peptides to cytotoxic T cells and may therefore enhance the anti-melanoma immune response.This vaccine induced a specific CTL response against B16F10 melanoma and showed strong protective as well as therapeutic effects (Zhu et al., 2010).
Another modification aiming in augmenting the presentation of Salmonella-delivered neuroblastoma antigens was proposed by the group of Lode (Fest et al., 2009;Huebener et al., 2008).The chimeric protein encoded by Salmonella plasmid contained a stably ubiquitinated tumor antigen, either tyrosine hydroxylase- (Huebener et al., 2008) or survivin (Fest et al., 2009) fragments.The presence of ubiquitin guaranteed proteasomal degradation of proteins encoded by the transgenes and increased MHC I-mediated peptides presentation.This approach led to significant CD8 + T cell-dependent inhibition of neuroblastoma growth (Fest et al., 2009;Huebener et al., 2008) and decreased the rate of metastasis (Huebener et al., 2008) or tumor growth upon rechallenge (Fest et al., 2009).
Manuel et al. also chose survivin as an excellent target for tumor therapy (Manuel et al., 2011).This antiapoptotic protein is expressed at a low level in normal adult tissues but is abundant in essentially all solid tumors (Altieri, 2003), as well as in endothelial cells during angiogenesis (Tran et al., 1999).However, unlike their predecessors, the researchers utilized bacterial promoter to express a codon-optimized transgene in Salmonella and used the SseF secretion signal to transport survivin to the cytoplasm of the infected cells.The combined therapy consisting of sequential intravenous injection of two Salmonella strains: the first encoding shRNA targeting a tolerogenic transcription factor, and the second coding for SseF-survivin, turned out to be effective even in the case of large, well-established melanoma tumors.The silencing of STAT3 expression, crucial for this very promising result, led to increased proliferation of intratumoral CD4 + and CD8 + T cells and elevated levels of granzyme B (Manuel et al., 2011).
Not only survivin but also other proteins overproduced in endothelium during angiogenesis, such as VEGFR2, one of the VEGF receptors, or endoglin, a component of the TGFβ (Transforming Growth Factor beta) receptor complex, have been chosen as targets for Salmonella-based vaccines (Jarosz et al., 2013;Jellbauer et al., 2012).The induction of the CTL response against these antigens should reduce angiogenesis and destroy tumor vasculature leading to inhibition of tumor development.Indeed, Salmonella carrying a plasmid encoding a fusion protein comprising the YopE secretion sequence and the fragment of VEGFR2 containing the CD8 + Tcell epitope efficiently limited the growth of B16F10 melanoma in both prophylactic and therapeutic settings (Jellbauer et al., 2012).The authors hypothesized that the vaccine may also affect the immunosuppresive tumor microenvironment by destroying the VEGFR2-positive subset of Treg lymphocytes.
Similarly, a combined therapy using orally applied Salmonella carrying endoglin cDNA with an intratumorally injected plasmid coding for interleukin-12 reduced microvessel density and diminished the number of Tregs within B16F10 tumors resulting in inhibition of the tu-mor growth and prolonged survival of mice (Jarosz et al., 2013).Some examples of in vivo studies on the efficacy of various Salmonella Typhimurium-based anti-cancer vaccines are presented in Table 4.
Recently, a phase 1 clinical trial evaluating the safety of the first Salmonella Typhi-based oral vaccine coding for the full length human VEGFR2 and aimed at eliciting an anti-VEGFR2 immune response has started and involves patients suffering from pancreatic cancer.The results of this trial are not yet available (Niethammer et al., 2012).

clINIcAl STudIeS INdIcATe TuMoR TARgeTINg AS The MAjoR lIMITATIoN of Salmonella-BASed TheRAPIeS
The mouse is a natural host for Salmonella enterica serovar Typhimurium and this species is considered to be the most sensitive to VNP20009 infections.However, the maximum tolerated dose (MTD) of VNP20009 in mice is large and was estimated to be 0.5 × 10 8 colony forming units per kg of body weight, which makes this attenuated strain at least 50 000 times less virulent than the parental Salmonella (Lee et al., 2000).Importantly, similar values of MTD were estimated for other species -dogs, pigs and monkeys (3.0 × 10 7 , 1.9 × 10 8 and 2.5 × 10 8 cfu/kg, respectively).Most of the adverse effects observed were transient and related to physiological responses to infection and stress, rather than to the intrinsic toxicity of the bacterial treatment.The studies in pigs found no endotoxic or septic shock reactions.In tumor-bearing mice, VNP20009 accumulates preferentially in tumors over livers at a ratio of 1000:1, but also in dogs with spontaneous neoplasia tumor colonization was detectable in 10 out of 24 Salmonella-treated cases (42%) (Thamm et al., 2005).Those promising features of preferential localization and apparent lack of toxicity allowed VNP20009 to enter clinical development in 1999.
In the first-in-man study performed by Vion Pharmaceuticals Inc., a total of 25 patients -24 with metastatic melanoma and one with metastatic renal cancer -were treated with 30-minute intravenous bolus infusions of VNP20009 dose ranging from 10 6 to 10 9 cfu/m 2 .On the basis of dose-limiting toxicity symptoms, the MTD of VNP20009 in humans was estimated to be 3.0×10 8 cfu/ m 2 .Adverse reactions at higher doses included fever, hypotension, anemia and thrombocytopenia and were probably due to cytokine release.Despite VNP20009 attenuation, significant amounts of TNF were detected in the patients' peripheral blood.Most adverse effects were mild and rapidly reversible, but none of the 25 patients experienced objective cancer regression.Bacteria were rapidly cleared from the peripheral blood and tumor colonization by Salmonella could be detected only in three patients (Toso et al., 2002).This result was clearly in contrast with data from the murine tumor models.
In a subsequent clinical study, additional 4 patients with metastatic melanoma received a 4-hour intravenous infusion of VNP20009 at the MTD of 3.0 × 10 8 cfu/m 2 .Adverse effects of the treatment included fever, chills and nausea, but were minor and transient.VNP20009 was detectable in samples of patients' blood up to 2 hours post treatment.No Salmonella could be cultured from tumor biopsies taken within 2 weeks of therapy (Heimann & Rosenberg, 2003).
The results of VNP20009 phase I clinical trials did not confirm Salmonella accumulation and tumor regres-U n c o r r e c t e d P a p e r i n P r e s s sion similar to previous preclinical data.However, the important finding is that VNP20009 can be safely administered to humans in large doses and that the toxicity of this bacterial strain is limited.Moreover, the first-inman study revealed that low tumor targeting in humans is a crucial therapeutic drawback of VNP20009.Since this feature can be significantly improved using genetic engineering, we made an attempt to enhance the accumulation of Salmonella in human tumors by constructing a VNP20009 strain expressing single chain antibody fragments specific to the carcinoembryonic antigen (CEA; Fig. 1).This VNP derivative was able to efficiently target CEA-expressing tumors in mice and could possibly overcome the targeting limitations of VNP20009 in humans, since CEA is present on more than 50% of human carcinomas (Bereta et al., 2007).
In summary, bacterial cancer therapy has recently moved beyond an interesting concept and has by now been supported by solid preclinical and clinical data.The unique features of bacterial therapeutics create the opportunity for novel anticancer strategies as bacteria meet all requirements for an ideal tumor-targeting agent.While the mechanism of action is often unclear, many examples have already proved that cancer treatment with bacteria can be site-specific, highly focused on the tumor and safe to other tissues.Among bacterial strains evaluated as cancer therapeutics so far, Salmonella Typhimurium is one of the most promising with first-in-man studies that support feasibility of clinical application.However, excellent tumor colonization observed in murine models was  not confirmed in humans, suggesting that tumor targeting is the major obstacle for further development.As many researchers focus on the effector molecules and other antitumor features of S. Typhimurium, insufficient localization in human tumors remains to be an unsolved issue.Targeting based on an antibody-fragment specific to TAAs is one of the few approaches proposed to overcome this drawback of therapeutic Salmonella strains.

*
figure 1. Tumor targeting of Salmonella via the anti-ceA antibody fragment.(A) a single chain antibody fragment (scFv) specific to carcinoembryonic antigen (CEA) was derived from a CEA-specific immunoglobulin G molecule; (B) A DNA sequence encoding scFv was fused with the gene coding for OmpA, an outer membrane protein of E. coli, and expressed in VNP20009, an attenuated strain of Salmonella Typhimurium.