Attenuated Salmonella typhimurium (S. typhimurium) strains can selectively grow and express exogenous genes in tumors for targeted therapy. We engineered S. typhimurium strain VNP20009 to secrete tumor necrosis factor-related apoptosis-inducing ligand (TRAIL) under the control of a hypoxia-induced nirB promoter and examined the efficacy of Salmonella-mediated targeted expression of TRAIL in mice bearing melanoma tumor and in TRAIL-resistant RM-1 tumor. We found that VNP preferentially accumulated in tumor tissues and the nirB promoter effectively drove targeted expression of TRAIL. Compared with recombinant TRAIL protein and VNP20009 combination therapy, VNP20009 expressing TRAIL significantly suppressed melanoma growth but failed to suppress RM-1 tumor growth. Furthermore, we confirmed that VNP20009 expressing TRAIL yielded its antitumor effect by inducing melanoma apoptosis. Our findings indicate that Salmonella-mediated tumor-targeted therapy with TRAIL could reduce tumor growth and extend host survival. (Cancer Sci 2012; 103: 325–333)
Tumor tissues consume a greater amount of oxygen than that supplied by the tumor-feeding blood vessels.(1) Normal tissues typically have median oxygen tensions of 40–70 mmHg, whereas that of most solid tumors is <10 mmHg. Clinically, poor oxygenation is a major indicator of adverse prognosis for solid tumors, which frequently contain large regions with low oxygen concentrations.(2–6) One contributing factor to the adverse outcome is the inadequate distribution of antitumor agents to the hypoxic areas of a tumor where blood flow is sluggish or highly irregular.(2) Strategies for improving tumor oxygenation have met with rather limited success.(7) Specific killing of hypoxic tumor cells by hypoxia-selective tumoricidal agents or modalities offers a separate strategy that views the presence of hypoxia in tumors not only as a determinant for poor prognosis but also as an opportunity for tumor-specific treatment.
Selective targeting of tumors where hypoxia exists enhances targeted concentration of therapeutic agents of interest in cancer cells while minimizing damage to the normal tissues, regardless of whether these therapeutic agents are chemotherapeutic drugs or recombinant proteins or ectopic genes. Attenuated Salmonella typhimurium (S. typhimurium) strains have been shown to selectively grow and express exogenous genes in tumors for targeted therapy.(8–12) We have previously shown that the facultative anaerobe S. typhimurium strain VNP20009 can replicate in hypoxic tumors(13–16) and is synergistic with cyclophosphamide against melanoma in a mouse tumor model.(17) Because the bacteria could preferentially accumulate in the hypoxic tumor microenvironment, there have been attempts to engineer controllable expression of therapeutic molecules such as prodrug-converting enzymes and antigens.(18–21) Tumor necrosis factor-related apoptosis-inducing ligand can selectively induce apoptosis in cancer cells and is effective against many cancers, including melanoma.(22)
Although S. typhimurium can thrive in the hypoxic environment of solid tumors after systemic infection, it can also survive in the oxygenated environment of normal tissues.(23,24) By surviving in normal tissues such as the liver, S. typhimurium engineered to secrete TRAIL may generate hepatotoxicity, as TRAIL has been found to induce hepatic cell death.(25) Therefore, selective accumulation of S. typhimurium in tumors not only enhances localized expression of therapeutic genes of interest but also avoids unintended toxicities from extra-tumor growth of S. typhimurium. The anaerobic-inducible nirB promoter can be heterologously expressed in facultative anaerobes and shows excellent stability,(26) and has been explored in Salmonella-mediated cancer immunotherapy.(27–33) Here, we engineered S. typhimurium strain VNP20009 to carry nirB promoter-driven vectors expressing our therapeutic gene of interest encoding TRAIL and we evaluated the distribution of VNP20009 and the effectiveness of the nirB promoter in mice. We further assessed the efficacy of VNP20009 engineered to produce TRAIL in a mouse melanoma model and mouse RM-1 prostate tumor. We found that VNP preferentially accumulated in tumor tissues and the nirB promoter effectively drove targeted expression of TRAIL. We also showed that Salmonella-mediated tumor-targeted TRAIL gene therapy could reduce tumor growth and extend host survival, suggesting the preclinical utility of Salmonella-mediated tumor-targeted therapy against melanoma.
Materials and Methods
Expression vectors. The prokaryotic pGFP expression vector (LacZ promoter, pLacZ-GFP) was purchased from Clonetech (Mountain View, CA, USA). The PCR primers and their sequences are listed in Table 1. sTRAIL was amplified from B6 mouse liver cDNA using primers TRAIL-F (nirB) and TRAIL-R. The amplified products were ligated with the nirB promoter by overlap PCR using primers nirB-F and TRAIL-R, which were subsequently cloned into the HindIII and SalI site of the plasmid pBR322 (Takara, Kusatsu, Japan) and the construct was designated pTRAIL. The gene encoding eGFP was amplified using primers GFP-F (nirB) and GFP-R, respectively, from the plasmid pEGFP (Clontech). The amplified products were then ligated to the nirB promoter and digested with HindIII and SalI, or directly digested with HindIII and SalI then subcloned into pBR322 and the construct was designated pNirBeGFP. The schematic of the plasmid constructs is shown in Figure 1. In addition, pET23a-TRAIL was constructed using primer TRAIL(pET)-F with NdeI and TRAIL(pET)-R with XhoI. Escherichia coli BL21 was used for protein expression and nickel affinity column was used for protein purification as described before.(34)
Table 1. The primer used in construction of the expression vectors
Primer sequence (5′–3′)
Forward primer of pnirB including HindIII enzyme site
Reverse primer of pnirB including overlaping sequence of GFP
Forward primer of GFP including overlaping sequence of pnirB
Forward primer of GFP including HindIII enzyme site
Reverse primer of GFPincluding SalI enzyme site
Reverse primer of pnirB including overlaping sequence of TRAIL
Forward primer of TRAIL including overlaping sequence of pnirB
Reverse primer of TRAIL including SalI enzyme site
Forward primer of TRAIL including NdeI enzyme site
Reverse primer of TRAIL including XhoI enzyme site
Bacteria, transformation, and colony formation assays. Lipid A-modified (msbB−) auxotrophic (purI−) S. typhimurium strain VNP20009 was obtained from ATCC (Rockville, MD, USA) and grown at 37°C to the mid-logarithmic phase in LB broth. The bacteria (2.0 × 108), harvested at 4°C, were dissolved in 40 μL 10% glycerol then mixed with 1 μL appropriate vectors (0.5 μg/μL) for electroporation using a Gene Pulser apparatus (Bio-Rad, Hercules, CA, USA) according to the manufacturer’s instructions and cultured on an LB plate with appropriate antibiotics at 37°C. For colony formation assays, the number of CFU was determined as previously described.(35) For induction of the expression of genes of interest, hypoxic or anaerobic LB medium was prepared by boiling to remove dissolved oxygen. Residual oxygen was driven out further by nitrogen for 15 min in boiling water. Then 100 μL 0.05% sodium sulfide and 0.05% cysteine were added to maintain the reducing environment in anaerobic jars (Oxid, London, UK). VNP20009 carrying appropriate expression vectors was inoculated and grown to the mid-logarithmic phase.
Annexin V/PI apoptosis. Purified msTRAIL and the lysis of VNPpTRAIL cultured under anaerobic conditions were incubated with B16F10 melanoma cells for 24 or 48 h. The cells were prepared and washed with binding buffer, then stained with annexin V–FITC for 15 min on ice in the dark. Then 1 μL PI was added to each sample for flow cytometry analysis.
Mouse antitumor studies. Six- to seven-week-old female C57BL/6 mice, purchased from the Laboratory Animal Center, Yangzhou University (Yangzhou, China), were housed in environmentally controlled conditions (22°C and a 12:12 h light:dark cycle, with the light cycle 06:00–18:00 and the dark cycle 18:00–06:00) with ad libitum access to standard laboratory chow and water. The study protocol was approved by local institution review boards and the animal study was carried out in accordance the ethical guidelines for animal use and care established by Nanjing University (Nanjing, China). B16F10 melanoma cells and RM-1 prostate cancer cells (ATCC) were grown as previously described.(17) The C57BL/6 mice were inoculated s.c. on the mid-right flank with 5 × 105 cells (B16F10) or 2 × 105 cells (RM-1) in 0.1 mL PBS. Mice were treated i.p. with S. typhimurium strain VNP20009 carrying appropriate plasmids at a dose of 105 CFU per mouse on day 7 (for B16F10) or day 9 (for RM-1) after tumor inoculation. Additionally, mice were injected i.p. with TRAIL protein at a daily dose of 100 μg per mouse for 10 days from day 3 after tumor inoculation. Tumor volume was determined using the formula: tumor volume = length × width2 × 0.52 and survival was recorded.
Western blot analysis. Tumor, liver, and spleen tissues were homogenized for 1 h on ice in 50 mM HEPES (pH 7.4), 100 mM NaCl, 50 mM NaF, 2 mM EDTA, 1% Triton-100, and 100 μg/mL PMSF and the tissue homogenates were subjected to the immunoblotting procedure as previously described.(36) Anti-TRAIL and anti-GAPDH antibodies (Santa Cruz Biotechnology, Santa Cruz, CA, USA) were used for immunoblotting studies. In addition, bacterial lysates were prepared from S. typhimurium strain VNP20009 or Escherichia coli BL21 carrying appropriate expression vectors for TRAIL and subjected to immunoblotting analysis as described above.
Immunohistochemistry, TUNEL assays, and fluorescence microscopy. Tissue sections (5 μm in thickness) were prepared according to standard protocols for H&E staining. Immunohistochemical studies using anti-Salmonella (Abcam, Cambridge, UK) anti-TRAIL (Santa Cruz Biotechnology) or its isotype control antibody were carried out as previously described.(37) In addition, TUNEL assays were carried out by detecting apoptotic nuclei using the TdT-FragEL DNA Fragmentation Detection Kit as instructed by the manufacturer (Merck, Darmstadt, Germany). For fluorescence microscopy, sections of frozen tumor tissues treated with VNP pNirBeGFP, VNP pLacZ-GFP, or control plasmids were prepared as above and examined under a fluorescence microscope and photographed. In addition, VNP20009 carrying pNirBeGFP or VNPpeGFP was grown to the mid-logarithmic phase under aerobic or anaerobic conditions. Aliquots of bacterial culture (1 mL) were harvested; 1 μL bacterial suspension was mixed with 10 μL deionized water on a glass coverslip and air dried for 20 min, then examined under a fluorescence microscope. The remaining bacterial suspension was lysed by sonication and filtered through a 0.22-μm pore-size filter to remove residual bacteria. The GFP intensity was read at 485 nm as previously described.(38)
Statistical analysis. Data were expressed as the mean ± SD and data analysis was carried out using spss software (SPSS 17.0, Chicago, IL, USA). Paired Student’s t-test analysis was carried out to assess statistical significance. Differences between experimental groups were considered significant if the P-value was < 0.05.
Expression of TRAIL in S. typhimurium effectively driven by nirB promoter. We designed plasmid constructs expressing genes of interest for Salmonella-mediated tumor-targeted therapy using the nirB promoter (Fig. 1). Our fluorescence microscopy revealed that the nirB promoter could effectively drive the expression of GFP in S. typhimurium strain VNP20009 carrying the plasmid construct pNirBeGFP under hypoxia (Fig. 2a,b), suggesting that the nirB promoter was a potent promoter. Immunoblotting studies further showed that TRAIL was effectively expressed in VNP20009 carrying the plasmid construct pTRAIL growing under hypoxia (Fig. 2c). These findings indicated that genes of interest for Salmonella-mediated tumor-targeted therapy could be effectively expressed in S. typhimurium strain VNP20009 carrying appropriate expression vectors driven by the nirB promoter.
Preferential accumulation of S. typhimurium strain VNP20009 in tumor tissues. We examined the in vivo distribution of S. typhimurium strain VNP20009 in B6 mice bearing melanoma. Colony formation assays showed that VNP20009 bearing pTRAIL could be recovered from tumor tissue, liver, and spleen (Fig. 3a), suggesting the distribution of VNP20009 to these tissue sites. Furthermore, on Amp+ or Amp− LB agar VNP20009 carrying pTRAIL yielded similar numbers of CFU in the tumor tissue, liver, or spleen, suggesting that the expression vectors can stably replicate without loss of VNP20009 in vivo, and also suggesting that the presence of appropriate expression vectors did not compromise the growth and proliferation of VNP20009. Comparison of the numbers of CFU in each target tissue site revealed that VNP20009 with appropriate expression vectors preferentially grew in the tumor tissues compared with the liver or spleen (P < 0.001). We further investigated whether our genes of interest were effectively expressed in B6 mice bearing melanoma. These findings indicated that S. typhimurium strain VNP20009 carrying appropriate expression vectors could be preferentially localized to tumor tissues, allowing the selective and effective nirB-driven expression in tumor tissues of genes of interest for Salmonella-mediated tumor-targeted therapy.
Next we examined the expression of TRAIL in the tumor, liver, and spleen tissues. We found markedly increased levels of TRAIL in the tumor tissues of mice inoculated with VNP20009 carrying pTRAIL but not in liver or spleen tissues (Fig. 3b). We also analyzed the distribution of Salmonella VNP20009 in tumor, liver, and spleen using anti-Salmonella antibody. We found VNP could only be detected in hypoxic and necrotic tumors but not in liver or spleen (Fig. 3c). Additionally, immunohistochemical study using anti-TRAIL antibody revealed that TRAIL was preferentially expressed in the hypoxic, necrotic area in the tumor, but TRAIL induction was not detected in oxygen-generating liver or spleen (Fig. 3c). These findings suggested that S. typhimurium strain VNP preferentially accumulated in tumor tissues, allowing localized expression of TRAIL driven by the nirB promoter.
Salmonella-mediated targeted therapy with TRAIL gene significantly suppressed melanoma growth. As we found that S. typhimurium strain VNP20009 could preferentially accumulate in tumor tissue and allow selective expression of genes of interest in the region for Salmonella-mediated tumor-targeted therapy, we evaluated the effect of VNPpTRAIL on tumor growth in vivo. We inoculated mice bearing melanoma with VNPpTRAIL or VNPpeGFP and also treated them with VNP20009 and purified msTRAIL protein (Fig. 4a). We found that VNPpTRAIL significantly inhibited melanoma growth with a markedly reduced tumor volume compared with controls (P < 0.001; Fig. 4b). We further compared the efficacy of VNPpTRAIL with the combination of recombinant TRAIL and VNP20009 (TRAIL + VNPpeGFP) in inhibiting melanoma growth in B6 mice. Although TRAIL + VNPpeGFP resulted in a marked inhibition of tumor growth compared with controls (P < 0.001), VNPpTRAIL was more effective than the combination in suppressing tumor growth, with a significantly smaller tumor volume. Moreover, msTRAIL was found to be expressed both in tumors treated with VNPpTRAIL and in tumors treated with TRAIL + VNPpeGFP. We found that VNPpTRAIL could lead to stronger msTRAIL expression in tumors than the combination therapy (Fig 4b). We also checked the in vitro sensitivity of melanoma for msTRAIL-induced apoptosis. We found that msTRAIL (Fig. 4c) could induce B16F10 melanoma apoptosis, and the lysis of cultured VNPpTRAIL could also induce melanoma apoptosis significantly compared with VNPpeGFP (Fig. 4d).
Salmonella-mediated targeted TRAIL gene therapy failed to suppress TRAIL-resistant RM-1 tumors. To verify whether TRAIL gene targeted therapy acts directly on tumor cells, we used reportedly TRAIL-resistant RM-1 prostate tumors.(39) We found that VNPpTRAIL targeted therapy could significantly suppress B16F10 melanoma tumors (P < 0.05) (Fig. 5a). However, VNPpTRAIL therapy failed to suppress TRAIL-resistant RM-1 tumor growth (Fig. 5b).
VNP20009 carrying pNirBeGFP could be specifically induced in hypoxic and necrotic tumor. To confirm the distribution of VNP20009 in mouse tissues and to verify the tumor-specific gene induction, we detected GFP expression in tumor, liver, and spleen by fluorescence microscopy. We found the VNPpLacZ-GFP (with constitutive promoter) could induce expression not only in necrotic tumor but also in spleen and liver tissue (Fig. 5c). However, VNPpNirBeGFP (with tumor specific promoter) could only be induced in hypoxic and necrotic tumor tissue (Fig. 5c).
Salmonella typhimurium strain VNP20009 carrying pTRAIL induced apoptosis of melanoma cells. VNP20009 could lead to tumor necrosis (Fig. 6a) and preferentially replicated in necrotic and hypoxic regions of tumors. We further examined possible mechanisms whereby VNPpTRAIL prolonged the survival of mice bearing melanoma. We carried out TUNEL assays of tumor, liver, and spleen samples on day 20 after tumor implantation. The assays showed that a greater number of tumor cells underwent apoptosis in mice receiving VNPpTRAIL compared with mice receiving VNPpeGFP (Fig. 6b), suggesting that pTRAIL expression in tumor could induce more melanoma cells to undergo apoptosis. In addition, no significant TUNEL-positive cells were detected in liver or spleen tissues of mice receiving VNPpTRAIL (Fig. 6b). Our findings indicated that VNPpTRAIL likely suppressed tumor growth and improved survival of melanoma-bearing mice by inducing apoptosis.
Some obligate or facultative anaerobic bacteria, such as Salmonella, preferentially multiply in hypoxic and necrotic areas of tumors and suppress their growth.(20,24,40,41) These so-called tumor-targeting bacteria have sparked interest in their use as antitumor agents and gene transfer vectors for tumor-targeted therapy.(20,41) Compared to conventional gene delivery carriers, bacterial vectors have some unique advantages, such as preferential colonization in the tumor with the ratio of bacterial titer in tumor to normal tissue of 1000–10 000:1. They also reside and replicate in the hypoxic regions of tumors for an extended period of time. These regions are considered resistant to chemotherapy and radiotherapy. The prolonged preferential presence of these bacterial carriers allows sustained expression of therapeutic genes of interests, thus exerting their tumoricidal or tumor inhibitory effects.(20,24,40,41)
Although S. typhimurium preferentially replicates in tumors, a small number of them may replicate in normal tissues, posing potential toxicity to normal tissues.(42) The gene products secreted by engineered S. typhimurium may also cause undesirable effects on normal tissues as a result of extra-tumor growth of the bacteria.(25) Therefore, S. typhimurium engineered to carry a promoter that allows preferential or specific expression of genes of interest could lead to enhanced localized expression of therapeutic genes of interest and at the same time avoid unintended toxicities to normal tissues. In the current study, we took advantage of the anaerobic-inducible E. coli nirB promoter(26,43,44) to induce the expression of genes of interest in the hypoxic environment of tumors. Fluorescence microscopy revealed that S. typhimurium was preferentially accumulated in the necrotic hypoxic areas of tumors. Our immunohistochemical study further indicated that TRAIL was preferentially expressed in the hypoxic, necrotic area of the tumor. Poor oxygenation of solid tumors predicts an adverse therapeutic outcome.(2–6,45–47) The selective growth of S. typhimurium VNP20009 in the necrotic hypoxic environment of melanoma tissues and the targeted expression of our genes of interest indicate that our strategy of using an anaerobic-inducible promoter for targeted tumor therapy offers a feasible and effective option for delivering therapeutic genes of interest selectively to the tumor tissues, particularly regions of hypoxia where resistance to adjuvant chemotherapy or radiotherapy is often present.
Our study also showed that nirB promoter-driven targeted expression of TRAIL from S. typhimurium in tumor tissue markedly suppressed the growth of melanoma in mice. More importantly, Salmonella-mediated tumor-targeted therapy with the TRAIL gene markedly extended the survival of mice bearing melanoma (data not shown). These findings indicate that Salmonella-mediated targeted therapy with the TRAIL gene driven by the anaerobic-inducible nirB promoter could offer a potentially effective therapeutic method for solid tumors such as melanoma, suggesting the preclinical utility of our study. Although we also showed the recombinant TRAIL protein could suppress tumor growth, the survival benefit was markedly less than that achieved with Salmonella-mediated targeted therapy with the TRAIL gene. Tumor-targeted TRAIL therapy by VNP20009 can lead to a more msTRAIL-enriched tumor microenvironment, leading to a more prolonged suppression effect than that achieved by TRAIL protein tumor therapy. Salmonella-mediated targeted therapy is also more convenient and cost-effective than repeated inoculations of the recombinant TRAIL protein of high purity.
It has been reported that TRAIL could activate the death receptor pathway and induce apoptosis of carcinoma cells.(48) Our TUNEL assays showed that Salmonella VNP20009 engineered to secrete TRAIL was associated with significant apoptotic activities of melanoma cells in mice, which could contribute to the tumor inhibitory effects and survival benefit observed in this study. For TRAIL-specific induction of tumor cell apoptosis, we used a TRAIL-resistant RM-1 tumor but only detected a poor tumor suppression effect compared with the VNP control. The results mean that our TRAIL gene delivery by VNP20009 could specifically lead to tumor cell apoptosis and contribute to its antitumor effect.
In conclusion, we devised a hypoxia-inducible nirB promoter-driven expression system to target poorly oxygenated tumors. We showed that Salmonella-mediated targeted therapy using this expression system leads to locally enhanced production of gene products of interest. We further showed that such therapy using the TRAIL gene could markedly extend survival in mice bearing melanomas. Our study provides a potentially effective method for treating recalcitrant tumors by specifically targeting poorly oxygenated tumor regions and inducing higher levels of tumor apoptosis.
The authors are grateful to grants from the Chinese National Natural Science Foundation (Grant Nos 30821006, 50973046, 31070706, 31071196, and 30973528), the Doctoral Station Science Foundation from the Chinese Ministry of Education (Grant No. 200802840023), the Jiangsu Provincial Natural Science Foundation (Grant Nos BK2010046, B22010074, BE2008639, and BY2009147), the National Drug Innovation Key Project (Grant No. 2009ZX09103-675), the National Key Basic Research Program (Grant No. 2011CB933502), the Bureau of Science and Technology of Changzhou (Grant Nos CS20092003, CQ20100009, CN20100016, and CZ20100008) and the Department of Science and Technology of Wujin District, Changzhou (Grant Nos WG2009007, WS201004).