Genetically Engineered Salmonella typhimurium as an Imageable Therapeutic Probe for Cancer

Attenuated Salmonella typhimurium reportedly has a propensity to naturally accumulate and replicate in a wide variety of solid tumors (1). For cancer therapy, attenuated S. typhimurium has been applied alone (2) or in combination with conventional therapeutics (3) and has also been manipulated to deliver therapeutic molecules (4). S. typhimurium bioengineered to generate bioluminescence (5, 6) or fluorescence (7) signals has been used to monitor bacterial migration to tumors in living small animal models, which may facilitate the prediction of the therapeutic efficacy associated with bacteriotherapy, through visualization of bacterial accumulation and replication in specific organs.

Here, we engineered attenuated S. typhimurium defective in guanosine 5′-diphosphate-3′-diphosphate (ppGpp) synthesis (ΔppGpp strain) to express both the cytotoxic protein cytolysin A (ClyA) to kill tumor cells and the bacterial luciferase gene lux to generate imaging signals. ClyA is a 34-kDa pore-forming hemolytic protein found in E. coli and Salmonella enterica serovars Typhi and Paratyphi A (8). A major problem for the full exploitation of cytotoxin-expressing Salmonella in cancer treatment is its toxicity in nontumoral reticuloendothelial organs, mainly the liver and spleen, due to their initial localization in these organs after tail vein injection (6, 9). Constitutive cytotoxin expression would thus inevitably result in hepatic or splenic injury. Therefore, for tumor-specific therapy with reduced toxicity in normal organs, it is necessary to use inducible promoter systems. Here, we used the P_BAD promoter from the E. coli arabinose operon, which can be activated by the sugar l-arabinose.

ΔppGpp S. typhimurium, SHJ2037 (relA::cat, spoT::kan), has been previously described (10). Bacterial doses were 4.5 × 10⁷ colony-forming units per mouse suspended in 100 μL of PBS.

The murine CT-26 colon carcinoma and Hep3B2.1-7 human hepatocellular carcinoma cell lines were obtained from the American Type Culture Collection (CRL-2638, HB-8064). The cells were grown in high-glucose DMEM containing 10% fetal bovine serum and 1% penicillin-streptomycin.

The expression plasmid pBAD-RLuc8 has been previously described (11). The clyA gene was amplified with 5′-AGTCCATGGTTATGACCGGAATATTTGC-3′ (forward primer) and 5′-GATGTTTAAACTCAGACGTCAGGAACCTC-3′ (reverse primer) using the S. typhi genomic DNA as a template (12). Amplified DNA was cut by NcoI and PmeI and used to directly replace RLuc8 at the same site in pBAD-RLuc8.

Bioluminescence imaging was done as previously described (13) using IVIS 100 (Caliper).

Five- to six-week-old male BALB/c and BALB/c athymic nu⁻/nu⁻ mice (20–30 g body weight) were purchased from the Orient Company (Korea). All animal care, experiments, and euthanasia were done in accordance with protocols approved by the Chonnam National University Animal Research Committee. Cells (10⁶) were harvested and suspended in 100 μL of PBS and injected s.c. into the right thigh of each mouse. Tumor volumes (mm³) were estimated with calipers every 4 d after tumor cell injection (6).

Homogenized samples, standardized for protein content at 180 μg, were subjected to SDS-PAGE on (12%) linear gradient gels and subsequently transferred to nitrocellulose membranes (Bio-Rad). To analyze ClyA, the membrane was probed with rabbit anti-ClyA antibody (1:250 dilution), horseradish peroxidase–conjugated goat anti-rabbit IgG (1:2,000 dilution), and luminol reagent (Santa Cruz).

Two-tailed Student's t tests were used to determine the statistical significance of the primary tumor growth difference between the control and treatment groups. In the case of metastasis scores, the statistical significance was determined by Mann-Whitney U tests. The survival analysis was done using the Kaplan-Meier curve and log-rank test. P < 0.05 was considered significant for all analyses. All data are expressed as mean ± SD.

To verify that remote control of the bacterial gene expression is possible, a bacterial expression plasmid was constructed in which a Renilla luciferase variant (RLuc8) was placed under the control of the P_BAD promoter (pBAD-RLuc8; Supplementary Fig. S1A). The mutant S. typhimurium ΔppGpp strain was electrotransformed with the pBAD-RLuc8 vector. Western blot analysis revealed that the intensity of RLuc8 expression (36.9 kDa) was directly proportional to the concentration of inducer, l-arabinose (0–0.4%), added to the cultured bacterial cells. RLuc8 was not detected in the absence of l-arabinose (Supplementary Fig. S1B and C). We i.v. administered the S. typhimurium ΔppGpp strain carrying pBAD-RLuc8 into immunocompetent BALB/c mice bearing CT-26 (n = 5). Because our previous observation (9) revealed that bacteria are cleared from the liver and spleen, and that they proliferate in tumors at 4 days post-inoculum (dpi), we induced RLuc8 expression with l-arabinose (60 mg) at 4 dpi. Bioluminescence was detected in the tumors after l-arabinose administration, but not in its absence (Supplementary Fig. S1D). These results strongly support the feasibility of using remote control of gene expression in tumor-targeting bacteria.

To control the therapeutic gene expression, the clyA gene was substituted for the RLuc8 gene in the pBAD-RLuc8 plasmid, generating the pBAD-ClyA plasmid (Fig. 1A). To noninvasively monitor the bacterial migration in mice, the bacterial luciferase (lux) operon from S. typhimurium-Xen26 (Xenogen-Caliper) was transduced into the SHJ2037 strain by P22HT int transduction (14). ClyA expression was proportional to the l-arabinose concentration (∼0.4%), and ClyA was not detected in its absence (Fig. 1B). Western blot analysis revealed the presence of ClyA protein (34 kDa) in both the pellet and the culture medium of the strain carrying pBAD-ClyA only after l-arabinose administration (Fig. 1C). This result indicates that ClyA protein expression was specifically induced and that the protein was secreted from the bacteria. The growth of salmonellae was not affected by ClyA induction (Fig. 1D).

To explore the in vivo antitumor activity of our engineered S. typhimurium, CT-26– or Hep3B2.1-7–bearing mice were treated with PBS and with untransformed or pBAD-ClyA–transformed S. typhimurium. Treatment of mice with transformed S. typhimurium in the presence of l-arabinose led to significant tumor shrinkage and, in some cases, tumor eradication (Fig. 2A and C). Survival was significantly prolonged in the group that received transformed S. typhimurium expressing ClyA by l-arabinose addition than in the other groups (Fig. 2D). Tumor shrinkage or the survival time following treatment with untransformed S. typhimurium or transformed S. typhimurium without induction was significantly better than that with PBS (control), but these tumors had a tendency to regrow. These results indicate that salmonellae alone are insufficient for significant tumor suppression.

Strong bioluminescence signals from the chromosomal lux operon of salmonellae were detected specifically in the tumor area of mice injected with bacteria (Fig. 2B). Most notably, the light signal from salmonellae diminished to an undetectable level in mice whose tumors were completely removed by the treatment (Fig. 2B,, bottom). This result implies that tumor-targeted salmonellae were cleared by the host immune system as the tumors were treated. Western blot analysis revealed the presence of ClyA protein in the tumor tissue only after l-arabinose induction (Fig. 2E). Histologic analyses revealed extensive central necrosis and remaining proliferative areas after i.v. injection of transformed S. typhimurium with induction of ClyA expression (Fig. 3A). ClyA protein was identified in areas between the necrotic and proliferative tumor regions (Fig. 3B). The observed body weight of treated mice was not significantly different from that of the control group, suggesting that expression of ClyA in tumors by i.v. administration of engineered Salmonella did not elicit explicit toxicity (Supplementary Fig. S2).

In a separate experiment, mice were injected i.v. with CT-26 cells expressing firefly luciferase to create lung metastases. Mice were i.v. injected with lux-expressing S. typhimurium 5 days after cancer cells were injected. The location of bacterial bioluminescence (Fig. 4A,, right) closely correlated with that of cellular (CT-26 FLuc) bioluminescence (Fig. 4A , left) in the lung fields, indicating that bacterial bioluminescence was detected specifically in the metastatic lung lesions. The specific location of bacteria in the lung lesions was further confirmed by imaging of isolated organs and gross necropsy (Supplementary Fig. S3).

The same protocol was then used to treat mice with lung metastases using pBAD-ClyA–expressing S. typhimurium. The effect of the treatment was evaluated either by weighing isolated lung masses or by a metastasis scoring method (15). Treatment with ClyA-expressing S. typhimurium markedly suppressed metastatic tumor growth better than control treatments (Fig. 4B–D; P < 0.01). Note that the effect of treatment with transformed S. typhimurium without ClyA induction was better than that with PBS (control; P < 0.05), indicating that the S. typhimurium ΔppGpp strain itself had some antitumor effect on the metastatic tumors.

In summary, these results provide proof of principle that (a) highly attenuated S. typhimurium ΔppGpp strain has an intrinsic tendency to target tumor tissue; (b) bacteria can be genetically engineered to deliberately express and secrete target protein into the tumor tissue; (c) the antitumor activity of our engineered bacteria was strong enough to suppress both primary and metastatic colon cancer in mice; (d) the P_BAD promoter is controllable and provides a very low background expression level (16); and (e) antitumoral bacteria labeled with imaging molecules allow determination of the bacterial fate in a simple, noninvasive manner that is amenable to repeated observations. Alternative approach has been exploited with strains of S. typhimurium that are dependent on the presence of leucine and arginine for growth. This strain was applied to treat many types of metastatic cancers in mouse models, and regression of tumor after i.v. administration was observed (17–19). To our knowledge, this is the first report of a successful bacterial engineering of an imageable therapeutic probe. Direct extensions of our current approach should allow the generation of multifunctional tumor-targeting bacteria, allowing safe, efficient, and individualized tumor therapy.

No potential conflicts of interest were disclosed.

Grant Support: Bio R&D program of the Korea Science and Engineering Foundation (KOSEF) funded by the Ministry of Education, Science, and Technology (MEST, 2008-04131); the National Research Foundation of Korea funded by the Ministry of Education, Science and Technology (no. 2009-0091729); the National R&D Program for Cancer Control, Ministry of Health and Welfare, Republic of Korea (0620330-1); and in part by the BioImaging Research Center at GIST. Y. Hong was supported by Regional Technology Innovation Program of the Ministry of Commerce, Industry, and Energy (grant no. RTI05-01-01); H.E. Choy was supported by a KOSEF grant funded by MEST (no. 2007-04213).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.