Tumor-targeting bacteria have been developed as powerful anticancer agents. Salmonella typhimurium VNP20009, a representative tumor-targeting strain, has been systemically administered as a single-agent therapy at doses of 1 × 106 to 3 × 106 colony-forming unit (cfu)/mouse, or in combination with other antitumor agents at doses of 1 × 104 to 2 × 105 cfu/mouse. Recently, we reported that oral delivery of VNP20009 at the dose of 1 × 109 cfu/mouse induced significant anticancer effects comparable to that induced by systemic administration of this strain at 1 × 104 cfu/mouse. To further address the efficacy and safety of oral administration of bacteria, here we performed a systemically comparative analysis of anticancer efficacy and toxicity of VNP20009 administered: (i) orally at a dose of 1 × 109 cfu/mouse (VNP9-oral); (ii) intraperitoneally at a dose of 1 × 104 cfu/mouse (VNP4-i.p.); or (iii) intraperitoneally at a dose of 1 × 106 cfu/mouse in tumor-free and tumor-bearing murine models. The results showed that VNP9-oral, similar to VNP4-i.p., induced significant tumor growth inhibition whereas VNP6-i.p. induced better anticancer effect in the B16F10 melanoma model. Among three treatments, VNP9-oral induced the mildest and reversible toxicity whereas VNP6-i.p. resulted in the most serious and irreversible toxicities when compared to other two treatments. Moreover, the combination of VNP9-oral with a low dose of chemotherapeutics produced comparable antitumor effects but displayed significantly reduced toxicity when compared to VNP6-i.p. The findings demonstrated that oral administration, as a novel avenue in the application of bacteria, is highly safe and effective. Moreover, the present preclinical study should facilitate the optimization of bacterial therapies with improved anticancer efficacy and reduced adverse effects in future clinical trials. (Cancer Sci 2009; 100: 2437–2443)
Tumor-targeting Salmonella can preferentially replicate in tumors and suppress their growth and metastasis in a variety of animal models.(1–5) These bacteria have been extensively investigated for anticancer therapies as a single-agent therapy,(3–10) tumor-targeted gene delivery vectors,(11–15) or in combination with other antitumor strategies, such as chemotherapy, radiotherapy, and angiogenesis inhibitor.(16–22) To ensure the success of bacterial infection, bacteria have been delivered systemically (intravenously [i.v.] or intraperitoneally [i.p.]) into animals and cancer patients.(2–21) Recently, we reported that tumor-targeting Salmonella could be administered orally as an effective anticancer agent.(23)
Salmonella typhimurium VNP20009 is a genetically attenuated anticancer strain carrying partial deletions in the purI and msbB genes.(24) As a frequently investigated strain, Salmonella typhimurium VNP20009 has been systemically administered for antitumor treatment as single-agent therapy with the doses of 1 × 106 to 3 × 106 colon-forming unit (cfu)/mouse,(8,24) or as part of combination therapy with the doses of 1 × 104 to 2 × 105 cfu/mouse,(16–18) or as a vector for tumor-targeted delivery of the pro-drug-activating enzyme with the dose of 1 × 106 to 3 × 106 cfu/mouse.(11–13) Our recent study showed that this strain could be delivered orally for effective anticancer therapy at the dose of 1 × 109 cfu/mouse which displayed a similar anticancer effect to that induced by systemic administration of 1 × 104cfu/mouse.(23) However, no systemically comparative evaluation in anticancer efficacy and potential toxicity of these treatments has been performed.
In this study, the high efficacy and safety of oral administration of bacteria were demonstrated by a systemically comparative analysis of anticancer efficacy and potential toxicity of VNP20009 administered: (i) orally at a dose of 1 × 109 cfu/mouse (VNP9-oral), an effective dose of orally delivered bacteria for cancer therapy;(23) (ii) intraperitoneally at a dose 1 × 104 cfu/mouse (VNP4-i.p.), a frequently used dose in combination therapy which displayed similar anticancer effects to VNP9-oral in the Lewis lung carcinoma model;(16,23) or (iii) intraperitoneally at a dose of 1 × 106 cfu/mouse, a frequently used dose for VNP20009-mediated single-agent therapy.(24) This study, using Salmonella typhimurium VNP20009 as a representative strain, provides preclinical data on the effect and toxicity of common modes of tumor-targeting Salmonella therapy, which could facilitate the optimization of bacterial therapies with improved anticancer efficacy and reduce adverse effects in future clinical trials.
Materials and Methods
Salmonella strain and bacterial infection. Lipid A-modified (msbB−), auxotrophic (purI−) Salmonella typhimurium VNP20009 (ATCC, Manassas, VA, USA) was cultured and prepared as described.(24) For intraperitoneal infection of VNP20009, 6- to 8-week-old female C57BL/6 mice were injected i.p. with 0.2 mL PBS containing 106 or 104 cfu bacteria. For oral administration of bacteria, mice were fasted for 3–4 h without food and drinking water, and then inoculated by gavage with 0.2 mL PBS containing 109 cfu bacteria, followed by feeding with adequate food and drinking water.(23)
Tumor implantation and evaluation of antitumor effects. B16F10 melanoma cells and Lewis lung carcinoma cells were purchased from the American Type Culture Collection (ATCC) and maintained in Dulbecco’s modified Eagle’s medium supplemented with 10% fetal bovine serum by routine culture methods. For tumor implantation, 6- to 8-week-old female C57BL/6 mice were implanted subcutaneously (s.c.) on the mid-right side with 5 × 105 B16F10 cells or 1 × 106 Lewis lung carcinoma cells in 0.1 mL PBS.(18,23) The antitumor activity of treatments was evaluated by tumor growth inhibition. Tumors were measured individually with a caliper. Tumor volumes were determined using the formula: tumor volume = length × width2 × 0.52.(17)
Evaluation of potential toxicity of bacterial infection. Six to 8-week-old female C57BL/6 tumor-free mice were randomly divided into four groups and treated by VNP6-i.p., VNP4-i.p., VNP9-oral, or PBS as a control. The potential toxicities induced by treatments were evaluated as indicated below.
Bacteria replication in normal tissues. Isolation and titration of bacteria from livers and spleens were performed as described.(23) Briefly, infected mice were euthanized and these tissues were removed aseptically and homogenized with PBS at a ratio of 5:1 (PBS volume [mL]: tumor weight [g]). Serially diluted homogenates were spread onto modified LB agar plates (without salt) and incubated at 37°C for 24 h. The titer of bacteria was determined by counting colonies and dividing them by the weight of the tissue (colony-forming unit [cfu]/g tissue).(23)
Flow cytometry analysis for the subpopulation of immune cells in spleen. Tumor-free mice were infected and spleens were collected post infection for subpopulation analysis of immune cells. Briefly, the spleens were ground and sieved to obtain the single cell suspension. The erythrocytes were lysed using lysis buffer before staining. Ten percent newborn cow serum in PBS was used as a blockage for 30 min. Immunofluorescent staining of cell surface markers was performed using FITC-F4/80 for macrophages, FITC-Gr-1 for granulocytes, PEcy5-CD4 for CD4 T cells, FITC-CD8 for CD8 T cells, and PE-B220 for B cells (BD, Franklin Lakes, NJ, USA). Samples were incubated in dark with shaking for 45 min. After washing in PBS, the cells were resuspended and then analyzed.
Hematoxylin–eosin staining of tissue sections. Livers were collected and weighted. Part of tissues were fixed with 4% formaldehyde, embedded in paraffin and sectioned for hematoxylin–eosin (H&E) staining according to standard histological procedures. Infectious foci in livers observed under a light microscope were used to evaluate the severity of inflammation response induced by bacterial infection.
Liver function evaluation. Serum samples were collected at 20 h, 10, 21, and 42 days post infection and stored at −80°C until serum alanine aminotransferase (ALT) and aspartate aminotransferase (AST) were measured using a Clinical Automatic Analyzer (Hitachi, Ibaraki, Japan).
Inflammatory cytokine analysis. Tumor-free mice were infected and serum samples were collected at 0 h, 2 h, 6 h, and 3 days post infection and stored at −80°C. The levels of tumor necrosis factor (TNF)-α, interleukin (IL)-6, and IL-1β were measured using commercial ELISA kits (JingMei, Shen Zhen, China).
Clinical signs and body weight. The clinical signs of infected animals were observed daily and body weight was measured at 6, 12, 18, and 28 days post infection.
Combination therapy with VNP9-oral and chemotherapeutics (cisplatin or 5-FU) when compared to VNP6-i.p. Six to 8-week-old female C57BL/6 mice were implanted with Lewis lung carcinoma cells as described above. After the tumors reached the size of about 0.1 cm3, mice were randomized into seven groups with seven mice per group. Group I received 2 mg/kg cisplatin in 0.2 mL of 0.9% NaCl solution injected i.p. once every other day for three times. Group II received comparable injections of 0.9% NaCl solution as a control. Group III were inoculated orally with VNP20009 at doses of 1 × 109(VNP9-oral) cfu/mouse once on the day 1 of treatment, and group IV were inoculated i.p. with bacteria at the doses of 1 × 106 (VNP6-i.p.) cfu/mouse. Group V was treated orally with 1 × 109 cfu Salmonella plus 2 mg/kg cisplatin administered as described above. Group VI received 20 mg/kg 5-FU in 0.2 mL of 0.9% NaCl solution injected i.p. once every other day for three times. The rest of group was treated orally with 1 × 109 cfu Salmonella plus 20 mg/kg 5-FU administered as group VI. The antitumor activity of treatments was evaluated by tumor growth inhibition and potential toxicity of treatment was evaluated by weight loss.
Statistical analysis. Results are reported as mean ± SD. Statistical significance of differences was assessed using SPSS software (version 10.0; SPSS, Chicago, IL, USA). anova followed by the least-significant-difference test was used for comparisons of tested parameters. The level of significance was set at P < 0.05.
Antitumor effect of bacterial treatments. Our previous study showed that VNP9-oral induced a similar anticancer effect when compared to VNP4-i.p. in the Lewis lung carcinoma model.(23) Here, we determined the antitumor effects of VNP9-oral, VNP4-i.p., and VNP6-i.p. in a B16F10 melanoma model. Consistently, VNP9-oral displayed significant antitumor efficacy comparable to that induced by VNP4-i.p., without significant toxicity to weight loss (Fig. 1a,b), whereas VNP6-i.p. led to better therapeutic effects but induced significant body weight loss (Fig. 1a,b). To test whether VNP9-oral in combination with low dose chemotherapeutics can produce a better anticancer effect with acceptable toxicity, VNP9-oral was combined with a low dose of cisplatin or 5-FU and compared to VNP6-i.p. with regard to anticancer effect and potential toxicity in the Lewis lung carcinoma model. Combination of VNP9-oral with cisplatin or 5-FU not only induced a significantly enhanced antitumor effect which was comparable to VNP6-i.p. (Fig. 1c,e), but also avoided significant weight loss coupled with VNP6-i.p. (Fig. 1d,f). These results suggested that combination therapy with orally delivered bacteria and other anticancer agents is a more desirable form than systemically administered single-agent bacteria therapy, such as VNP6-i.p.
Potential toxic effects of bacterial treatment on the spleen. Although Salmonella VNP20009 preferentially replicated in tumor, it multiplied in normal tissues, especially in the spleen and liver, at considerable levels.(8,25) To evaluate the toxicity of different bacterial treatments on the spleen, the levels of bacteria replication in spleens were first determined at 20 h, 10, 20, and 42 days post bacterial treatments. As shown in Figure 2, a higher amount of bacteria was found in spleens infected with VNP6-i.p. than in those infected with VNP9-oral and VNP4-i.p. at all examined time points. Compared with VNP6-i.p and VNP4-i.p., VNP9-oral harbored the lowest amount of bacteria in spleens. Consistent with bacterial levels in spleens, VNP6-i.p. induced more severe splenomegalia than VNP9-oral and VNP4-i.p., while VNP9-oral induced the mildest increase in spleen weight among the three infections (Fig. 2).
Effect of bacteria infection on the subpopulation of immune cells in the spleen. As one of the most important immune organs, the spleen harbors a group of active immune cells and plays an essential role in the regulation of immune response. To further evaluate the effects of bacterial treatments on the spleen, we determined the percentage of macrophages, granulocytes, CD4+ T cells, CD8+ T cells, and B cells in spleens by flow cytometry assay using immunostaining for specific molecular markers of these immune cells. As shown in Figure 3, bacterial infections induced a significant increase in macrophages and granulocytes, but a decrease in CD4+ T cells, CD8+ T cells, and B cells in spleens. Again, VNP6-i.p. led to the most significant changes in these parameters when compared to VNP9-oral and VNP4-i.p which resulted in modest changes in these parameters.
Bacterial replication and inflammation induction in the liver. We then determined bacterial replication in the liver. As shown in Figure 4(a), VNP6-i.p. resulted in higher bacterial replication in this tissue than VNP9-oral and VNP4-i.p. during the whole observation period. We next quantified the infection foci within livers to evaluate the severity of tissue inflammation response to bacterial infection. Similarly, the number of infection foci in VNP6-i.p. was significantly higher than VNP4-i.p. and VNP9-oral at all time points (Fig. 4b,c). Among the three infection regimens, VNP9-oral led to the mildest bacterial replication, especially at 20 h and 10 days post infection (Fig. 4a), and inflammatory response (Fig. 4b,c).
Effects of bacterial treatments on liver function. In the liver, an inflammatory response following bacterial infection may lead to dysfunction. Here, we measured the level of serum transaminases of mice at 20 h, 10, 20, and 42 days post bacterial infection. As shown in Figure 5, bacterial infection induced a significant increase in ALT and AST. Among the three treatments, VNP6-i.p. resulted in the rapidest increase of AST and ALT with peak titers at about 20 h post treatment, whereas VNP4-i.p. and VNP9-oral did not result in any increase at this time point. At subsequent time points, the level of AST and ALT induced by VNP6-i.p. were also significantly higher than those induced by VNP4-i.p. and VNP9-oral. Consistent with bacteria replication and inflammatory foci in livers, VNP9-oral led to the most modest increase in AST and ALT when compared to the other two treatments (Fig. 5).
Induction of inflammatory cytokines. Bacterial infections induced the production of inflammatory cytokines including TNF-α, IL-6, and IL-1β which are classical markers for the evaluation of potential toxicity of bacterial treatments. Here, we found that bacterial infection induced the production of TNF-α, IL-6, and IL-1β with peak titers at 2 h post infection (Fig. 6). Consistent to the severity of bacterial replication and inflammatory response, VNP6-i.p. induced the most significant increase in these inflammatory cytokines whereas VNP4-i.p. and VNP9-oral induced a modest increase or no change when compared to control treatment (Fig. 6).
The effects of bacterial infection on weight gain and clinical signs. Body weight and clinical signs are major markers for the evaluation of treatment-related toxicity in preclinical studies. Here, we determined the effects of bacterial infection on these parameters. We found that VNP6-i.p. induced significant weight loss (Fig. 7) and apparent clinical signs, including lethargy and emaciation (data not shown), in mice soon after infection and these adverse effects lasted the whole observation period. In sharp contrast, VNP4-i.p. and VNP9-oral had no obvious influence on weight gain (Fig. 7) and infected mice looked as normal as the controls (data not shown). Similar results were obtained from B16F10 melanoma-bearing mice and Lewis lung carcinoma-bearing mice following these three treatments (data not shown).
Salmonella typhimurium VNP20009 has been systemically administered as a single-agent therapy at doses of over 1 × 106 cfu/mouse, or as a part of combination therapy at doses of over 1 × 104 cfu/mouse in preclinical studies.(8,16–18) Here, we report that oral delivery at a higher dose of 1 × 109 cfu/mouse could also induce promising therapeutic effect, which was comparable to VNP6-i.p. when combined with chemotherapeutics. Oral delivery is a relatively novel strategy for bacterial cancer therapy in which therapeutic bacteria is formulated as an oral preparation in further clinical research. Since toxic adverse effects caused by bacteria administration is the main hindrance to further clinical application, we performed this comparative study on the potential toxicity of these forms of bacterial administration, and based on the findings, hope to develop treatment strategies with a balance between antitumor efficacy and toxicity. Compared to systemic infection of bacteria (VNP4-i.p. and VNP6-i.p.), VNP9-oral exhibited higher safety in all tested parameters, including bacteria accumulation within tissues, histopathological response, liver function (serum AST and ALT), inflammatory cytokine induction, and weight loss, as well as clinical symptoms. The high safety of oral delivery of bacteria should directly result from its invasion routes. Unlike systemic injection (i.p. or i.v.) of Salmonella which led to multiple tissue infection quickly and severely, most of the bacteria delivered orally were killed or drained from the host, and only a small percentage of inoculum penetrated the wall of gut and infiltrated into the host, therefore leading to significantly less toxicity.(26,27)
The high safety of the oral administration of bacteria may broaden the application of tumor-targeting Salmonella as an anticancer agent. It is well established that oral treatment of non-tumor-targeting S. typhimurium containing DNA vaccine could induce specific anticancer immunity.(28,29) We found that oral delivery of recombinant tumor-targeting S. typhimurium VNP20009 could induce a specific immune response against a reporter antigen gene hemagglutinin (unpublished data), suggesting VNP20009 as a potential vaccine delivery vector. Using tumor-targeting Salmonella as a delivery vector for anticancer DNA vaccine, dual therapeutic effects, mediated by both specific antitumor immunity and bacteria therapy, would be induced.
Most recently, it was reported that humeral immune response against Salmonella (anti-Salmonella antibody) in immunized mice significantly inhibited tumor-targeting potential, gene transfer capability, and anticancer effects of systemically administered tumor-targeting Salmonella.(30) The interference of circulating anti-Salmonella antibody existing in animals or cancer patients may pose a big challenge to systemic administration of bacteria for effective anticancer therapy. In contrast, orally delivered tumor-targeting bacteria may directly penetrate into the tumor tissues of oral, gastric, or colorectal carcinomas, and avoid being eliminated by blood-circulating antibacterial antibodies. Therefore, oral delivery of bacteria may be more suitable than systemic infection for the treatment of these types of cancers in antibacterial antibody-carrying patients.
Systemic administration of VNP20009 at doses of 1 × 106 to 3 × 106 cfu/mouse has been widely used for single-agent anticancer therapy in previous studies. Our present study clearly demonstrated that VNP6-i.p. resulted in significant systemic toxicity although it did not kill infected mice (data not shown), which suggested that VNP6-i.p. was not applicable as a single-agent therapy. However, combination of VNP9-oral with low dose cisplatin or 5-FU produced comparable antitumor effects but displayed less toxicity when compared to VNP6-i.p. Similar results were also obtained by combining VNP9-oral with cyclophosphamide, another classical chemotherapeutic, in our previous study.(23) The synergic antitumor effect of combination therapies would result from the fact that such chemotherapeutics could create a more suitable microenvironment, such as more severe tumor hypoxia and necrosis formation, for bacteria to target and replicate.(21,22) The findings demonstrated that combination therapies with bacteria in safer modes (such as VNP9-oral) and other anticancer agents could achieve the best benefits for Salmonella-mediated therapy with regard to anticancer effects and potential toxicity.
In summary, the present systemically comparative study on the anticancer effect and potential toxicity of common modes of Salmonella therapy strengthens the feasibility and applicability of oral administration of tumor-targeting bacteria for anticancer therapy. Although Salmonella VNP20009 was used in this study, other tumor-targeting Salmonella strains are probably also applicable to oral administration. Thus, oral administration may provide an alternative route for the safe delivery of bacteria for effective anticancer therapy.
The authors are grateful to the Chinese Nature Science Foundation (nos. 30425009, 30500637, 30730030) and National Science Fund for Creative Research Groups (no. 30821006), the Doctoral Station Science Foundation from the Chinese Ministry of Education (no. 200802840023), the Key Project of Jiangsu Provincial Nature Science Foundation (no. BK2007715), Jiangsu Province’s Key Medical Center Project, and the Key Project from Jiangsu Provincial Department of Health (no. H200524).