Optimized Doxycycline-Inducible Gene Expression System for Genetic Programming of Tumor-Targeting Bacteria

To improve the expression balance between P_tetA and P_tetR and the transfer frequency of the plasmids (Fig. 1), we modified pJL39 to generate pJH18 in which TetR was controlled by the weak constitutive promoter P_araB, and bidirectional tet promoters, P_tetA and P_tetR, could be used for simultaneous expression of two cargo genes (Fig. 1A). Furthermore, bom sequence was also cloned in this plasmid for its ability to facilitate bacterial maintenance in antibiotic-free conditions (Fig. 1B). Using pJH18, we constructed four plasmids in which a reporter gene (rluc8 for Renilla luciferase variant 8) and/or a therapeutic gene (clyA for cytolysin A) were placed under the control of P_tetA or P_tetR; pJH18-RR encoding rluc8 under P_tetR, pJH18-AR encoding rluc8 under P_tetA, pJH18-CR encoding clyA and rluc8 under P_tetA and P_tetR, respectively, and pJH18-RC encoding rluc8 and clyA under P_tetA and P_tetR, respectively (Fig. 1A).

We transformed ∆ppGpp Salmonella with the pJH18-CR (SAM-CR) or pJL87 (SAM-pJL87) plasmid encoding clyA and rluc8 under P_tetA and P_tetR, respectively [8]. TetR expression and bom activity were compared between the two transformed bacteria. In SAM-CR, TetR proteins (23 kDa) were constitutively expressed in every growth time of Doxy-free condition (Fig. 2A), whereas in SAM-pJL87, TetR was undetectable at any growth time point. We note that TetR expression relied on Doxy in pJL87, as it is expressed by P_tetR [8]. To assess plasmid maintenance in SAM-pJL87 (without bom) and SAM-CR (with or without bom), these transformants were spread onto agar plates after overnight culture in the absence of ampicillin. After 24 h of culture, SAM-CR (with bom) showed no significant change of colony numbers in the presence or absence of ampicillin, whereas only 12.9 % of the SAM-CR (without bom) and 1.7 % of the SAM-pJL87 colonies remained in the presence of ampicillin compared with those in the absence (Fig. 2B). This result indicates that pJH18 is stably maintained for 24 h, even without antibiotics because it contains bom sequence.

Next, we evaluated Doxy-inducible expression of a reporter gene in cultured SAM transformed with pJH18-RR (SAM-RR) or pJH18-AR (SAM-AR). Doxy-induced Rluc8 (36.9 kDa) expression was measured in bacterial cultures after 4-h induction with various Doxy concentrations (Fig. 3). The luciferase activity by Rluc8 was specifically identified only in the presence of Doxy. Doxy-induced Rluc8 expression could be detected as a saturated amount in western blot analysis even at the lowest Doxy concentration (10 ng/ml) (Fig. 3A). The Rluc8 protein level was 2- to 6-fold higher under P_tetA than under P_tetR. The western blot result was consistent with that of a luciferase activity assay (Fig. 3B), with luciferase activity being detected even with 10 ng/ml of Doxy and being saturated at 20 ng/ml of Doxy. This luciferase activity was 3-fold higher with P_tetA than with P_tetR (Fig. 3B).

Subsequently, we engineered bacteria carrying pJH18-RC (SAM-RC) or pJH18-CR (SAM-CR) and assessed cargo gene expression. In western blot analyses, both Rluc8 (36.9 kDa) and ClyA (34 kDa) proteins could be specifically identified only in the presence of Doxy (Fig. 4A). In SAM-RC and SAM-CR, the expression levels of these proteins were 2- to 6-fold higher under P_tetA than under P_tetR. This result was consistent with Rluc8 enzyme activity assays (Fig. 4B). The ClyA activity secreted from SAM-RC or SAM-CR was also evaluated using a blood agar plate; SAM-RC and SAM-CR could lyse blood cells only after Doxy induction (Fig. 4C).

We then compared the cargo expression levels of SAM-CR and SAM-pJL87, which express ClyA and Rluc8 under P_tetA and P_tetR, respectively (Fig. 5). After Doxy induction, the expression of Rluc8 and ClyA was approximately 80-fold and 5-fold stronger in SAM-CR than in SAM-pJL87, respectively (Fig. 5A). In particular, Rluc8 expression under the control of P_tetR showed a remarkable increase in SAM-CR. The expression difference was also reproduced in Rluc8 activity tests. Rluc8 luciferase activity was 30- to 90-fold higher in SAM-CR than in SAM-pJL87, depending on the concentration of Doxy (Fig. 5B). After Doxy induction, SAM-CR showed stronger hemolytic activity than SAM-pJL87 (Fig. 5C). It should be noted that SAM-CR grew well, similar to non-transformed bacteria, despite the strong expression of ClyA and Rluc8 after Doxy induction (Fig. 5D).

The arabinose-inducible expression system using P_BAD promoter has been widely used for precise and strong induction of protein expression in bacteria-based cancer therapy [9, 27]. Thus, we assessed cargo expression in SAM transformed with pJH18 or pBAD (Fig. S1). The ClyA and Rluc8 expressions driven by P_BAD were comparable to those by P_tetA and P_tetR. The expression of ClyA and Rluc8 proteins in SAM-pBAD-ClyA was 1.4-fold higher than in SAM-RC (P_tetR::clyA) but 1.3-fold lower than in SAM-CR (P_tetA::clyA) (Fig. S1A). This result was consistent with the Rluc8 activity assay (Fig. S1B). Furthermore, we also found that in the single-cassette-polycistronic pBAD system encoding ClyA and Rluc8 (pBAD-CR (P_BAD::clyA::rluc8), Fig. 1A), there was no significant difference in proximal ClyA expression between SAM-pBAD-CR and SAM-CR, but the expression of Rluc8 protein was 3.2-fold lower in SAM-pBAD-CR than that in SAM-CR (Fig. S2A). This result was corresponded with the bioluminescence activity (Fig. S2B). The result indicates that the pJH18 system in the present study is appropriate for balanced expression of two genes.

In our previous study, we demonstrated Rluc8 activity driven by P_tetR in tumor tissues when SAM-pJL87 was injected into mouse tumor models, but its level was 100-fold lower than that driven by P_tetA promoter [8]. We evaluated the improvement in the efficiency of gene expression under P_tetR in SAM-CR by comparing it with that of SAM-pJL87 in CT26 xenograft tumors (Fig. 6). Mice were injected with SAM-CR or SAM-pJL87 via the tail vein and were orally administered Doxy (17 mg/kg body weight) at 3 days postinoculum (dpi) (Fig. 6A). Rluc8 activity (driven by P_tetR) in CT26 xenografts treated with SAM-CR was 14-fold higher than in those treated with SAM-pJL87 (Fig. 6 B and C), and ClyA protein expression (driven by P_tetA) was 3-fold higher in tumor tissues treated with SAM-CR than in those treated with SAM-pJL87 (Fig. 6D). These results indicate that the novel Doxy-inducible system pJH18 was able to express dual genes in tumor-targeting bacteria much more efficiently than was pJL87.

Interestingly, there were no significant differences in bacterial numbers in the mouse tumors treated with SAM-CR until 5 dpi, regardless of Doxy induction (Fig. 6E). By contrast, bacterial numbers in mouse tumors treated with SAM-pJL87 decreased more than 10-fold after Doxy induction compared with numbers before Doxy induction at 3.5, 4, and 5 dpi. We speculate that Doxy induction affects plasmid maintenance in the pJL87 system and that bom sequence in pJH18 is critical to prevent plasmid loss during bacterial growth in tumor tissue.

To further assess the antitumor activity of ClyA expressing bacteria, C57BL/6 mice bearing B16F10 were treated with PBS, attenuated S. typhimurium harboring empty vector pJH18 (SAM-E) or pJH18-CR (SAM-CR) through intravenous injection, respectively. Mice revealed a decrease in body weight after bacterial treatment compared with PBS treatment, which might be due to inflammation after bacterial treatment (Fig. S3A). The best therapeutic effect was monitored in mice treated with SAM-CR, and the moderate suppression was observed in mice treated with SAM-E (Fig. S3B and C). These data demonstrated that engineered bacteria secreting ClyA triggered antitumor activity.

We finally evaluated whether in vivo gene expression in the SAM-CR strain could be enhanced by a lower dose of Doxy inducer. When induced with a 10-fold lower dose of Doxy (1.7 mg/kg body weight), bioluminescence was strongly detected in tumors colonized by SAM-CR but not in those colonized by SAM-pJL87 (10-fold higher in the former than in the latter) (Fig. S4A and B). This result indicates that gene expression could be enhanced by transforming bacteria with the novel Doxy-inducible pJH18 system.

E. coli DH5a-competent cells were purchased from Enzynomics (Daejeon, Korea) and used for all gene cloning work. The SAM bacterium was a ΔppGpp-defective S. typhimurium SHJ2037 (relA::cat, spoT::kan) strain that is used to measure gene expression and activity in various plasmids [36]. All bacteria were grown in Luria-Bertani (LB) broth at 37 °C with agitation.

A murine colon carcinoma CT26 cell line was purchased from the American Type Culture Collection (MD, USA). The cells were cultured in high-glucose Dulbecco’s Modified Eagle Medium (DMEM) (Welgene, Gyeongsangbuk-do, Korea) containing 10 % fetal bovine serum (FBS) and 1 % penicillin–streptomycin and were incubated at 37 °C in an incubator with 5 % CO₂.

pJH18, a plasmid with novel bidirectional tet promoter (P_tetA and P_tetR), was engineered as follows. Using pJL39 plasmid as a template [8, 37], the tetR OFR [37] was amplified by polymerase chain reaction (PCR) with a TetR-R(EcoRI) and TetR-F(XbaI) primer set. After digestion with EcoRI and XbaI, the fragment was introduced into the same enzyme sites of pBAD plasmid and named pBAD-TetR [9]. The plasmid has tetR ORF driven by P_BAD promoter. The fragment containing a bidirectional tet promoter and multiple cloning sites at both ends (MCSI and MCSII) was amplified by PCR using a Tet-F(AflII) and Tet-R(HindIII) primer set with pJL39 as a template. The amplified fragment was digested with AflII and HindIII and inserted into the same enzyme sites of pBAD-TetR, which was called pTet-BAD. The bom fragment was amplified by the bom-F(PciI) and bom-R(NheI) primer set using pBAD-Rluc8 as a template and was cloned between PciI and NheI enzyme sites of pTet-BAD after enzyme digestion. It was named pTetII which has lost P_BAD promoter. Finally, a constitutive promoter P_araB was amplified by a P_araB-F(MfeI) and P_araB-R(MfeI) primer set using E. coli DH5α genomic DNA as a template and was cloned upstream of tetR ORF in pTetII using the Gibson assembly method (New England Biolabs, MA, USA). The resulting plasmid was named pJH18 and used it as a backbone in this study.

Various cargo gene fragments were amplified by PCR. The rluc8 gene was driven by P_tetA promoter (P_tetA::rluc8) and was amplified by an Rluc8/ClyA-P_tetA-F(StuI) and Rluc8-P_tetA-R(SpeI) primer set using the pBAD-Rluc8 plasmid as a template. The rluc8 gene driven by P_tetR promoter (P_tetR::rluc8) was amplified by an Rluc8-P_tetR-F(KpnI) and Rluc8-P_tetR-R(HindIII) primer set using the pBAD-Rluc8 plasmid as a template.

The clyA gene driven by P_tetA promoter (P_tetA::clyA) was amplified by an Rluc8/ClyA-P_tetA-F(StuI) and ClyA-P_tetA-R(SacI) primer set using the pBAD-ClyA plasmid as a template. The clyA driven by P_tetR promoter (P_tetR::clyA) was amplified by a ClyA-P_tetR-F(KpnI) and ClyA- P_tetR-R (SalI) primer set using the pBAD-ClyA plasmid as a template.

The PCR fragments were digested by the indicated restriction enzymes and cloned into the corresponding sites in pJH18. The resulting plasmids were named pJH18-RR (P_tetR::rluc8), pJH18-AR (P_tetA::rluc8), pJH18-CR (P_tetA::clyA, P_tetR::rluc8), and pJH18-RC (P_tetA::rluc8, P_tetA::clyA).

To evaluate the expression level of distal genes/ORFs, we developed pBAD-ClyA-Rluc8 from pBAD-ClyA system. The rluc8 gene was amplified by an Rluc8-P_BAD-F(SalI) and Rluc8-P_BAD-F(SphI) primer set using the pBAD-Rluc8 plasmid as template. The PCR products were digested by the indicated restriction enzymes and cloned into the corresponding sites in pBAD-ClyA for generating pBAD-ClyA-Rluc8 (P_{BAD::clyA::cluc8}).

The primers and plasmids used in cloning were shown in Tables S1 and S2 and Fig. 1.

For western blot analysis, SAM bacteria transformed with the plasmids were incubated overnight in LB medium containing ampicillin (100 μg/ml). The cultured bacteria were inoculated into a medium with 1:100 dilutions and cultured. At an optical density of 0.6 at 600 nm (OD₆₀₀), Doxy [0–500 ng/ml in ethanol (w/v)] was added, and the bacteria were further cultured. After 4 h, the bacteria were precipitated with centrifugation at 4000 rpm for 5 min. After simple washing with phosphate-buffered saline (PBS), the bacterial pellets were dissolved in sodium dodecyl sulfate (SDS) buffer containing 0.2 % β-mercaptoethanol (Sigma-Aldrich, Darmstadt, Germany) and were heat-treated for 5 min at 95 °C. The dissolved bacterial proteins (0.5 OD₆₀₀ equivalents per lane) were separated in 15 % SDS-PAGE (40 % Acrylamide, Tris-HCl (pH 6.8)/Tris base (pH 8.8), SDS, ammonium persulfate (APS), TEDMED) in running buffer (Tris base, glycine, sodium dodecyl sulfate) and then transferred to nitrocellulose membrane (GE Healthcare, MA, USA) by transfer buffer (Tris base, glycine). The membranes were incubated in blocking buffer [5 % (w/v) skim milk in Tris-buffered saline buffer (Tris base, NaCl) containing 0.1 % Tween 20 (Sigma-Aldrich, Darmstadt, Germany) (TBS-T)] at room temperature for 1 h. After decanting the blocking buffer (TBS + 5 % w/v skim milk), primary antibodies in 1X TBS with 1 % skim milk were added to the membranes, and they were incubated for 2 h. After decanting the primary antibodies and simple washing with TBS-T (500 ml TBS 1X + 500 μl Tween) (three times), secondary antibodies conjugated with horseradish peroxidase (HRP) were added to the membranes and incubated for 1 h. Before adding the chemiluminescence HRP substrate (Merck Millipore, MA, USA), the membranes were washed with TBS-T three times. The specific protein bands against each antibody were detected with a ChemiDoc^TM XRS+ system imager (BIO-RAD, California, USA). All antibodies used in this study are listed in Table S3.

For the luciferase activity assay, SAM bacteria transformed with the pJH18 plasmids containing the Rluc8 gene were induced by increasing the concentration of Doxy in the early exponential phase (OD₆₀₀, 0.5~0.7) for 4 h. The bacterial pellets (OD₆₀₀, 0.5 per sample) were resuspended in 100 μl of PBS to measure Rluc8 activity immediately after addition of 5 μl of coelenterazine (0.2 μg/μl) at room temperature. Light emission was measured using a Microlumat Plus LB96V luminometer (Berthold Technologies, Bad Wildbad, Germany) or a Lumina S5 in vivo imaging system (IVIS; Perkin Elmer, MA, USA) with an exposure every 1 s. The measured values were calculated as relative luminescence units [38].

Female 6-week-old BALB/c mice and C57BL/6 (~20 g in body weight) were purchased from Orient (Seongnam, Korea) and keep in specific pathogen-free conditions of mouse room for 1 week prior to starting any experiments. The mice were housed in a group of 3–5 mice in per plexiglass cage (22.3 × 26.8 × 13.0 cm) under 12h/12h at light/dark cycle (light on at 7:00 am) in a temperature-controlled room (22– 24 °C) and 40–60 % humidity, with free access to tap water and standard chow diet (PMI Nutrition International, LLC 4001 Lexington Avenue North, Arden Hills, MN 55126). Cages were lined with full autoclave wood fiber (Northeastern Product Corp., Warrensburg, NY, 12885) and replaced by clean cages 2 times/week. No enrichment material was used inside the cage. At the moment of the experiments, mice were weighing approximately 18 g and aging 7 weeks. The animals were randomly selected from different cages to house them in one new cage for tumor implantation, and each mouse was used for experiment only once. All experimental processes were performed from 11:00 to 7:00 pm. After culture in DMEM media, CT26 cells (10⁶ cells per mouse) or B16F10 cells (5 × 10⁵ cells per mouse) were detached and subcutaneously injected into mice anesthetized with 2 % isoflurane. After cell injection, it took about 12 days for the tumor volume to reach 150–180 mm³. All animal care and experimental procedures were performed following the guidelines of the Animal Care and Use Committee, Chonnam National University (Gwangju, Korea), the National Centre of the Replacement, Refinement and Reduction on Animals in Research [39].

For bacterial injections, SAM transformants cultured overnight were inoculated into fresh LB media containing ampicillin in a 100-fold dilution ratio and were further incubated until an OD₆₀₀ of 2~2.5 (early stationary phase) was attained. After centrifugation at 4000 rpm for 5 min, bacterial pellets were washed with PBS twice. Bacteria (3 × 10⁷ CFU/100 μl in PBS) were intravenously injected into the mice via the tail vein. To induce cargo genes in SAM transformants, the indicated amounts of Doxy (Sigma-Aldrich, MO, USA) were orally administered using gavage (starting at 3-day postinoculum, once a day after 1 h of fasting) [8]. Tumor volume (mm³) was calculated using the formula (length × height × width)/2 of the tumor in millimeters [8]. Most data analysis and experiments were blinded to prevent any bias.

To obtain bioluminescence images 12 h after oral Doxy administration, tumor-bearing mice were anesthetized with 2 % isoflurane and then received coelenterazine (0.7 mg/kg body weight) through intravenous injection. After 1 min, bioluminescence imaging was acquired from the mice using an IVIS system, as described previously [17, 40].

To measure the existence of cargo molecules, tumor tissues were obtained from mice and homogenized in protein extraction solution (20 mM Tris-Cl pH 6.8, 150 mM NaCl, 5 mM EDTA, 1 % NP-40, protease inhibitor 1X (Xpert protease inhibitor cocktail solution 100X, GenDEPOT)). The homogenized tissue samples were centrifuged and filtered through a 0.2-μm filter. A total of 20 μg of proteins were separated using 15 % SDS-PAGE, and western blot analysis was performed with specific antibodies.

After sacrificing the mice, tumors and other organs were imaged with IVIS. Tumors were homogenized and filtered as the sample preparation for the western blot analysis. The filtrates were serially diluted with PBS and spread out on LB agar plates with or without ampicillin. After overnight culture, the bacteria were enumerated, and the results were expressed as CFU/g of organ.

Statistical analysis was performed using GraphPad Prism 8.0 software with p < 0.05 considered significant. Two independent groups were compared using unpaired two-tailed t-tests. Comparison of multiple experimental groups was evaluated using one- or two-way ANOVA with multiple comparisons post hoc test to obtain multiplicity-adjusted P value. All data are shown as mean ± SEM.