Attenuated Salmonella engineered to produce human cytokine LIGHT inhibit tumor growth

Protein Expression and Intratumoral Accumulation of Attenuated Salmonella.

To engineer the attenuated S. typhimurium strain purI⁻/msbB⁻ for expressing LIGHT, we used the plasmid pGEN206, in which a human LIGHT cDNA was expressed with an N-terminal appended leader sequence that directs protein secretion under control of an ompC promoter. The engineered bacteria were grown in culture and recovered by centrifugation, and the resulting cell-containing pellet and cleared culture supernatant were tested for LIGHT protein by immunoblotting, revealing the presence of LIGHT in both cells and supernatant (Fig. 1A). Control Salmonella (transformed with empty plasmid) and their culture supernatants did not contain LIGHT protein.

Salmonella reportedly accumulate in solid tumors in vivo (4, 5, 20). To explore the tumor targeting of the attenuated purI⁻/msbB⁻ strain, we transformed these bacteria with a plasmid encoding luciferase and injected them i.v. into tumor-bearing mice, using a bioluminescence whole animal imaging method to localize the bacteria. In five of five mice evaluated, strong luminescent emissions were detected specifically in s.c. tumors (see Fig. 1B for example). We conclude therefore that the attenuated S. typhimurium strain used here retains its ability to target tumors in vivo. Furthermore, this capability remains unchanged after transformation with empty or LIGHT-bearing plasmids, respectively [supporting information (SI) Fig. 4]. In addition, for all six of the carcinoma lines tested, namely CT-26, D2F2, B16, LLC, HeLa, and PPC-1, attenuated Salmonella were able to infect the cancer cells in culture as revealed by the presence of GFP-emitting bodies in the cytosol (data not shown).

Salmonella Expressing LIGHT Inhibit Primary and Metastatic Tumor Growth.

We explored the antitumor activity of LIGHT-expressing Salmonella in immunocompetent BALB/c mice bearing s.c. syngeneic tumors, using either D2F2 breast (Fig. 2A) or CT-26 colon (Fig. 2B) carcinoma cells. Mice with established tumors (≈50 mm³) were injected i.v. with PBS, untransformed Salmonella, Salmonella transformed with empty plasmid, or Salmonella expressing LIGHT. Marked reduction of tumor growth was observed in mice treated with Salmonella expressing LIGHT, compared with control bacteria and PBS-treated mice (Fig. 2 A and B). Salmonella containing empty plasmid were not significantly different from PBS, indicating that bacterial targeting to tumors by itself is insufficient for antitumor activity.

To evaluate the antitumor activity of LIGHT-expressing bacteria in a more stringent model of cancer, we injected CT-26 cells i.v. to establish pulmonary metastases, and then treated mice 9, 14, and 19 days later with PBS, untransformed Salmonella, bacteria transformed with empty plasmid, or LIGHT-expressing Salmonella. Using either lung weights (Fig. 2C) or a metastasis scoring method (Fig. 2 D and E) to evaluate effects, LIGHT-expressing bacteria markedly suppressed metastatic tumor growth, compared with control-treated animals. Again, untransformed Salmonella and Salmonella containing empty plasmid were not significantly different from PBS, indicating that bacterial targeting to tumors by itself is insufficient for antitumor activity.

During our experiments, no animals from control or treatment groups died, lost weight, or showed overt signs of disease. To more carefully determine whether LIGHT-expressing Salmonella cause toxicity, we performed histological analysis of tissues and organs of BALB/c mice bearing s.c. tumors at 34 days after i.v. injection with LIGHT-expressing Salmonella. Lungs and kidney were completely normal. Liver showed mild inflammation, as predicted by prior studies (5), presumably because of bacteria uptake by Kupffer cells. The major histopathological finding was splenomegaly [416 ± 61 mg for Salmonella, 451 ± 101 mg for Salmonella plus empty vector and 429 ± 82 mg for Salmonella plus LIGHT vector vs. 165 ± 35 mg for PBS-treated mice (n = 3)], which was characterized by expansion of both the white and the red pulp. Analysis of circulating hematopoietic cells showed slight lymphocytosis in some instances, but was otherwise normal (data not shown). Thus, treatment with LIGHT-expressing Salmonella is well tolerated, particularly for a cytokine-based cancer therapy.

Antitumor Activity of LIGHT–Expressing Bacteria Is Dependent on LIGHT Receptors.

The mechanism accounting for the antitumor activity of LIGHT-expressing Salmonella was investigated by using neutralizing soluble receptor-Ig (R-Ig) fusion proteins that target the LIGHT receptors HVEM and LTβR (21, 22). For these experiments, C57BL/6 mice were injected s.c. with Lewis lung carcinoma cells and the neutralizing HVEM-Ig or LTβR-Ig. As with the other tumor models, LIGHT-expressing Salmonella but not control Salmonella greatly inhibited tumor growth (Fig. 3A). Anti-HVEM and anti-LTβR partially neutralized the antitumor activity of LIGHT-expressing Salmonella (P < 0.05).

Histological examination of tumors by hematoxylin/eosin (H&E) staining showed massive inflammatory cell infiltrates in tumors obtained from mice treated with LIGHT-expressing bacteria, whereas few leukocytes were seen in the tumors of PBS-treated mice, and only slight infiltrates were detected in mice treated with untransformed Salmonella or with Salmonella containing empty vector (SI Fig. 5). Flow cytometry analysis of cell suspensions prepared from resected tumors showed striking increases in B cells (CD19⁺) and in both CD4⁺ and CD8⁺ T cells (Fig. 3B). Depleting CD4 or CD8-positive cells in vivo by using complement-fixing anti-CD4 or anti-CD8 antibodies neutralized most of the antitumor activity of LIGHT-expressing Salmonella, suggesting that both CD4⁺ and CD8⁺ T cells make contributions to the antitumor activity. Immunodepletion using an antibody targeting the NK cell antigen NK1.1 also reduced the antitumor activity of LIGHT-expressing Salmonella (data not shown), implying a role also for these immune effector cells.

Spleens from mice receiving various treatments showed an approximate doubling of DCs [336 ± 48% versus 164 ± 27% CD11c⁺CD205⁺ cells (compared with PBS control); Fig. 3C] for mice treated with Salmonella containing LIGHT or control plasmid, respectively, versus empty plasmid (P < 0.05). Thus, consistent with the documented role of LIGHT as a DC cell growth factor (12), LIGHT-expressing Salmonella induced increases in DC cells in vivo.

The massive inflammatory cell infiltration into tumors of mice treated with LIGHT-expressing bacteria suggested involvement of chemoattractants (SI Fig. 5). LIGHT is known to induce expression of chemokine CXCL9 (23), which leads to attraction of CXCR3 chemokine receptor-expressing cells, such as T cells (24). Using an enzyme-linked immunosorbent assay (ELISA), we determined that levels of intratumoral CXCL9 were approximately doubled in mice treated with LIGHT-expressing Salmonella compared with control bacteria (Fig. 3D). Thus, the antitumor activity of LIGHT-expressing Salmonella is associated with increased expression of CXCL9.

Animals, Bacterial Strains, Cell Lines, and Plasmid Vectors.

Female BALB/c and C57BL/6 mice were purchased at ≈6–8 weeks of age from The Jackson Laboratory (Bar Harbor, ME). All animal experiments were performed according to the National Institutes of Health Guide for Care and Use of Experimental Animals and approved by the Animal Care Committee of the Burnham Institute for Medical Research (AUF 04-152). Attenuated lipid A-negative S. typhimurium (purI-/msbB⁻) were generated as described (4). The D2F2 cell line was a gift from W.-Z. Wei (Karmanos Cancer Institute, Wayne State University, Detroit, MI); CT-26 and Lewis lung carcinoma cells were purchased from the American Type Culture Collection (ATCC, Manassas, VA). cDNAs encoding luciferase (pGL3; Promega Corporation, Madison, WI), enhanced green fluorescent protein (pEGFP; Clontech, Mountain View, CA), or an N-terminal secretory leader sequence (MISSSSIS; described in ref. 26), followed by human LIGHT (InvivoGen, San Diego, CA), were inserted into pGEN206. This vector is a slightly modified form of pGEN222 (25) in which the parA locus has been exchanged with a parM and a parR locus. In addition, a repA locus has been introduced between the origin of replication and the multiple cloning sites. Protein expression of pGEN206-LIGHT was demonstrated by immunoblotting of 10⁹ cfu Salmonella, bearing nothing or vector, or their respective culture supernatants with a polyclonal rabbit anti-human LIGHT antibody provided by C.F. Ware (La Jolla Institute for Allergy and Immunology, La Jolla, CA). The experiment to demonstrate infection of tumor cells via Salmonella in vitro was performed by a standard gentamycin protection assay (26).

Optical Luminescence Imaging of Mice and Image Processing.

Live animal luminescence optical imaging was performed 15 min after i.p. injection of 10 mg of luciferin per mouse by using the Xenogen system (Hopkinton, MA). Imaging parameters were selected and implemented by using the instrument's Image Pro Plus 5.1 software. Bright field and luminescence photographs were obtained and merged with the same program. The reported images are real-time unprocessed images.

Tumor Cell Challenge and Treatment with Salmonella and Antibodies.

For experiments involving primary tumors, BALB/c mice (n = 5–8 per group) were challenged s.c. with a lethal dose of either 1.25 × 10⁵ D2F2 breast carcinoma cells or 2.5 × 10⁵ CT-26 colon carcinoma cells, then treated with either PBS or 5 × 10⁶ cfu of various types of Salmonella per mouse at days 9, 14, and 19, as indicated. Tumor volumes were measured by using an external caliper at various times and calculated as length/2 × width². To establish pulmonary metastases, BALB/c mice (n = 8) were first injected i.v. with 5 × 10⁴ D2F2 breast carcinoma cells and then treated i.v. with PBS, or indicated Salmonella (5 × 10⁶ cfu per mouse) after 6, 13, and 20 days. After 50 days, lungs were weighed (mean ± SEM) and examined for metastases, assessing the percentage of lung surface covered by metastases as follows: 0, 0%; 1, <20%; 2, 20–50%; 3, >50%. For experiments involving antibody-mediated depletion/blocking, C57BL/6 mice (n = 5) were challenged s.c. with 4 × 10⁵ Lewis lung carcinoma cells and treated i.v. after 7 days with PBS, Salmonella, Salmonella plus empty vector, or Salmonella plus LIGHT-bearing vector, including groups of mice that have been treated i.p. with 500 μg per mouse of antibodies against CD4⁺ T cells (anti-CD4 clone GK1.5; ATCC), CD8+ T cells (anti-CD8 clone 2.43; ATCC), NK cells (anti-asialo-GM; Wako Chemical, Dallas, TX), or with 100 μg per mouse of neutralizing soluble receptor-Ig (R-Ig) against HVEM (HVEM-Ig), or LTβR (LTβR-Ig, both provided by Biogen Idec, San Diego, CA) after 5 and 15 days. Animals were killed and tumor weights were measured at 22 days.

Immunohistochemistry, FACS Analysis, and ELISA of Tumors and/or Spleens.

Mice were killed with an overdose of anesthetic. For immunohistochemistry, tissues were fixed in paraformaldehyde, embedded in paraffin, and stained with H&E. For additional details, see SI Experimental Procedures. For FACS analysis, tissues from CT-26 tumor-bearing mice (n = 3) treated as outlined above were dispersed into single cell suspensions and stained according to the manufacturer's suggestions, by using individual and combinations of antibodies directly conjugated with fluorochromes (BD, Franklin Lakes, NJ), including stains for DCs (CD11c⁺/CD205), NK cells (DX5⁺), B cells (CD19⁺), CD4⁺ T cells (CD3⁺/CD4⁺), or CD8⁺ T cells (CD3⁺/CD8⁺). The results shown consist of comparisons of the different groups with the PBS control (= 100% ± SD). For ELISA, a portion of tumors from the same mice (n = 3) was extracted by using a 1% Nonidet P-40 solution in PBS, and analyzed by ELISA for the CXCL9 chemokine (R & D Systems, Minneapolis, MN). Data represent nanograms of cytokine per gram of total protein.

Statistics.

Statistical significance of data from control and treatment groups was determined by two-tailed Student's t test. The significance of metastases scores was determined by Mann–Whitney U test. Findings were regarded as significant, if P < 0.05.