Genetic Programming by Nitric Oxide-Sensing Gene Switch System in Tumor-Targeting Bacteria

Recent progress in synthetic biology has enabled bacteria to respond to specific disease signals to perform diagnostic and/or therapeutic tasks. Salmonella enterica subsp. enterica serovar Typhimurium (S. Typhimurium) colonization of tumors results in increases in nitric oxide (NO) levels, suggesting that NO may act as a candidate inducer of tumor-specific gene expression. The present study describes a NO-sensing gene switch system for triggering tumor-specific gene expression in an attenuated strain of S. Typhimurium. The genetic circuit was designed to sense NO via NorR, thus initiating the expression of FimE DNA recombinase. This was found to lead sequentially to the unidirectional inversion of a promoter region (fimS), which induced the expression of target genes. Target gene expression in bacteria transformed with the NO-sensing switch system was triggered in the presence of a chemical source of NO, diethylenetriamine/nitric oxide (DETA/NO) in vitro. In vivo results revealed that the gene expression is tumor-targeted, and specific to NO generated by inducible nitric oxide synthase (iNOS) after S. Typhimurium colonization. These results showed that NO was a promising inducer to finely tune the expression of target genes carried by tumor-targeting bacteria.


Introduction
Synthetic biology is a multidiscipline that artificially engineers genetic circuits to display specific biological behaviors in cells [1][2][3]. These genetically engineered bacterial cells act as novel therapeutics, with advantages when compared with small molecules or biologic agents. The localization of these cells can be better controlled, their retention times can be prolonged, and their ability to release a sufficient dose of drug in response to specific biomarkers can be better regulated [4]. Several microorganisms have been engineered for diagnostic or therapeutic purpose. Engineered Escherichia coli (E. coli) have been developed to sense reactive oxygen species (ROS) indicator, tetrathionate, to detect gut inflammation [5], and engineered Salmonella enterica subsp. enterica serovar Typhimurium (S. Typhimurium) were developed to deliver anticancer proteins to tumor tissues, with genes placed under the control of promoters that respond to specific signals [3].
Since streptococcal microorganisms were first injected into cancer patients to suppress malignant tumors in the 19th century [6], the anticancer activities of various bacteria have been explored [7]. S. Typhimurium has shown high tumor-targeting specificity, the ability to penetrate deep into tumor tissues, and robust anticancer activity. An attenuated S. at 37 • C. Transformed bacterial cells were stored at −80 • C in 25% glycerol before in vitro and in vivo assessments of the NO-sensing gene switch system.

Construction of Plasmids for NO-Sensing Gene Switch System
NIA3, a plasmid containing the NO-sensing elements (P norV , fimE, P lacIq -NorR, and fimS), was the kind gift of Dr. M. Shapiro (California Institute of Technology, Pasadena, CA, USA). The plasmids pTD103-sfGFP (#48885, Addgene) and pTU2S-a (#74088, Addgene), containing fragments of the super-folder green fluorescent protein (sfGFP) gene and the p15A origin of replication (ori), respectively, were gifts from Dr. J. Hasty (University of California, San Diego, CA, USA) and Dr. P. Freemont (Imperial College London, UK), respectively. pBAD-Rluc8 with pBR322 ori was described previously [35,36]. The gene fragments were amplified with PCR using KOD Plus Neo DNA polymerase.
The P norV fragment in the NIA3 plasmid was amplified using the primers P norV -F (5 -CG GAA TAT ATC CCC TAG GGC GCT GAA AAC GAT CCT G-3 ) and P norV -R (5 -TAG CAA CCT CAA TTT ATT CAG CGT GTT C-3 ). The fimE fragment lacking a ribosomal binding site (RBS) in the NIA3 plasmid was amplified using the primers noRBS-FimE-F (5 -GAA TAA ATT GAG GTT GCT AAT GGT GAG TAA ACG TCG TTA TCT TAC C-3 ) and FimE-R (5 -ATC CGC CGC CCT AGA CCT AGG GAC TAG TTC AAA CCT CTT CTC TTT TTA ATT TTT C-3 ). The two PCR fragments were ligated using a Gibson assembly kit, with the resulting P norV ::fimE fragment ligated into the AvrII-cut pTU2S-a plasmid. The plasmid containing p15A ori, called pFimE, contained a fimE gene driven by the P norV promoter and a chloramphenicol resistance (Cm R ) gene.
The fimS (IRL-fimS-IRR) fragment in the NIA3 plasmid was amplified using the primers fimS block-F (5 -AAA AAC ATG TCG CAG CCA TGA CCC AGT C-3 ) and fimS block-R (5 -GGT ACC TTT CTC CTC TTT AAT ACA AGA ACA ATT GGG GCC CAT TTT G-3 ). The rluc8 fragment encoding Renilla luciferase variant 8 with a strong RBS sequence (AAAGAGGAGAAA) in the plasmid pBAD-Rluc8 was amplified using the primers Rluc8-F (5 -AAA GAG GAG AAA GGT ACC ATG GCT TCC AAG GTG TAC GAC-3 ) and Rluc8-R (5 -TGC GGT CGA CCT GCT CGT TCT TCA GCA CG-3 ). These two fragments were ligated by Gibson assembly kit and named fimS::rluc8. The fimS::rluc8 fragment and pBR-NOR were digested with SalI and PciI and ligated by T4 DNA ligase. The resulting plasmid was named pSRluc8.
The sfGFP gene fragment with a strong RBS sequence in the plasmid pTD103-sfGFP was amplified using the primers sfGFP-F (5 -AAA GAG GAG AAA GGT ACC ATG AGC-3 ) and sfGFP-R (5 -AGC TTT GTT TAA ACT CAT TTG TAG AGC TCA TCC ATG CCA  TG-3 ), with this fragment ligated with the above-described fimS fragment using a Gibson assembly kit. This fragment, named fimS::sfGFP, was digested with PciI and PmeI, and ligated by T4 DNA ligase into pBR-NOR digested with same restriction enzymes. The resulting plasmid was named pSGFP.
All plasmids were sequenced (Macrogen, Korea). The resulted plasmids and SL. ∆ppGpp transformants were listed in Table S1.

Analysis of Nitric Oxide (NO)-Sensing Gene Switching Events
SL. pNASR and SL. pRluc8 were cultured at 37 • C overnight with shaking at 200 rpm in LB broth supplemented with ampicillin (100 µg/mL) and/or chloramphenicol (34 µg/mL). These cultures were diluted 1:100 into 3 mL of fresh LB broth and further cultured to mid-log phase, defined as an optical density at 600 nm (OD 600 ) of 0.5. The bacterial cultures were treated with various concentrations (0-1000 µM) of diethylenetriamine/nitric oxide (DETA/NO, Abcam, Cambridge, UK) and incubated with shaking at 37 • C. After 16 h, the bacterial pellets were harvested by centrifugation and washed with phosphate-buffered saline (PBS).
To measure the Rluc8 activity of the activated switch, 10 µL aliquots of 40 µg/mL coelenterazine, a luciferase substrate, were added to pellets resuspended in 200 µL PBS in a 96-well plate. The bioluminescence signals were measured with OrionL Microplate luminometer (Titertek Berthold, Pforzheim, Germany) and normalized to OD 600 . The bioluminescence images of each well were taken with an IVIS Lumina imaging system (PerkinElmer, Waltham, MA, USA). The levels of expression of Rluc8 protein were determined by western blotting using primary anti-Renilla luciferase antibody (Abcam, Cambridge, UK) and secondary horseradish peroxidase (HRP)-conjugated antibody. DnaK was used as control and detected with anti-DnaK antibody (Enzo Life Science, Farmingdale, NY, USA).
To measure the activated switch percentage under various concentrations of DETA/NO, SL. pNASGFP were treated with DETA/NO for 16 h, washed, and resuspended in PBS. The bacterial cells were serially diluted with PBS and spread onto LB agar plates containing antibiotics. After culture for overnight, the plates were placed under UV light to count green-fluorescent colonies. The percentages of green-fluorescent colonies relative to total colonies were calculated.

PCR Analysis of Switch Inversion
SL. pNASR were treated with the indicated concentrations of DETA/NO, and 100 µL aliquots were killed at 100 • C for 5 min. The switch events of fimS were detected by PCR using the primers, fimS (5 -GTC GAG CCA CAG AAA CGT TAG CTT TAC ATA TAG-3 ) and rluc8 (5 -TCA CTG CTC GTT CTT CAG CAC GC-3 ), with the expected size of the PCR band of the switched fimS::rluc8 gene being~1.1 kb. Each reaction contained 1 µL of each primer (10 pmol) and 1 µL of heat-killed bacterial cells in a total reaction volume of 20 µL, with the amplification protocol consisting of 30 cycles.

Measurement of Time-Dependent Switch Activation
SL. pNASR were treated with 100 µM DETA/NO and further cultured at 37 • C. Aliquots were collected after 0-8 h in culture, with the resulting bacterial pellets resuspended in PBS. Rluc8 activity was measured by determining bioluminescence intensities in the presence of coelenterazine, as described above. The switch events were also analyzed by PCR using fimS and rluc8 primers.

Switch Inheritance Verification
SL. pNASR were cultured in the presence of 100 µM DETA/NO for 16 h, diluted 100-fold in 3 mL of fresh media with antibiotics and without DETA/NO, and further cultured for 72 h. Rluc8 activity at the indicated times was measured by determining bioluminescence intensities in the presence of coelenterazine.

Analysis of In Vivo Switch Activation via Bioluminescence Imaging of Rluc8 Expression
All animal experiments and euthanasia procedures were performed in accordance with protocols approved by the Animal Research Committee of Chonnam National University, Korea. Female Balb/C mice aged 6 weeks were purchased from Orient (Korea) and anesthetized with 2% isoflurane (Hana Pharm, Seoul, Korea). Tumors were generated by subcutaneous implantation of 5 × 10 5 CT-26 colon carcinoma cells (CRL-2638, American Type Culture Collection, Manassas, VA, USA) in 100 µL PBS into the right flank of each mouse. When the tumor sizes reached about 150 mm 3 , SL. pNASR (1 × 10 7 in 100 µL PBS) were injected intravenously into each mouse. The mice were subjected to in vivo bioluminescence imaging with coelenterazine (20 µg in 100 µL PBS, i.v.) on the indicated days after bacterial injection using an IVIS Lumina S5 imaging system (PerkinElmer, Waltham, MA, USA). The levels of bioluminescence in the tumor region were measured as corresponding photon signals.

In Vivo NO-Specific Switch Activation via Inducible Nitric Oxide Synthase (iNOS) Activity Inhibition
To assess the effects of inhibition of iNOS activity, an iNOS inhibitor, 1400 W dihydrochloride (200 µg per injection), was intraperitoneally injected into CT-26 tumor-bearing mice every day, beginning 3 days before bacterial injection to the end of the experiment. Bioluminescence images were obtained on the indicated days after SL. pNASR injection. The levels of bioluminescence in the tumor region were measured as corresponding photon signals.

Measurement of iNOS Expression in Tumors Treated with Bacteria Bearing the NO-Sensing Gene Switch System
Mice treated with SL. pNASR or PBS (control) were sacrificed on the indicated days. The tumors were removed, weighed, and homogenized in a 4-fold volume of ice-cold NP40 lysis buffer containing 1x protease inhibitor cocktail (genDEPOT, Barker, TX, USA). After incubation for 1 h on ice, the samples were centrifuged at 13,000 rpm for 30 min at 4 • C and passed through a 0.2 µm filter, with the supernatants stored at −80 • C prior to further analysis. The supernatants were thawed on ice, and their total protein concentrations were measured using a BCA assay kit (Thermo Fisher Scientific, Waltham, MA, USA). The protein concentration of each sample was adjusted to 1 µg/µL with NP40 lysis buffer containing 1x protease inhibitor cocktail. Aliquots containing 10 µg protein were separated on 8% SDS-PAGE gels and transferred to PVDF membranes (GE Healthcare Life Sciences, Chicago, IL, USA). The membranes were incubated overnight at 4°C with mouse anti-iNOS monoclonal antibody (1:1000, Abcam, Cambridge, UK) or rabbit anti-β-actin antibody (1:1000, SantaCruz, Dallas, TX, USA), followed by incubation with HRP-conjugated anti-mouse Ig (1:1000, Invitrogen, USA) or anti-rabbit Ig (1:500, Thermo Fisher Scientific, Waltham, MA, USA) secondary antibodies for 1 h at room temperature. The resulting bands were photographed with ChemiDocTM XRS+ machine (Bio-Rad, Des Planies, IL, USA). The intensities of iNOS bands were measured by ImageJ software and normalized against those of β-actin bands.

Immunofluorescence Staining of sfGFP in Tumors Treated with Bacteria Carrying the NO-Sensing Gene Switch System
Mice treated with SL. pNASGFP or PBS (control) were sacrificed on the indicated days after bacterial injection. Tumor tissues were embedded in optimal cutting temperature (OCT) compound (Thermo Fisher Scientific, Waltham, MA, USA) and stored at −80 • C. Frozen tumor sections (5 µm thick) were made with a cryomicrotome (Thermo Fisher Scientific, Waltham, MA, USA) and mounted onto glass slides. The slides were blocked with 5 % BSA at room temperature for 1 h, and then incubated overnight at 4 • C with mouse anti-iNOS monoclonal antibody (1:100, Invitrogen, Waltham, MA, USA). After three washes with PBS containing 0.5% Tween-20 (PBS-T), the slides were incubated with Alexa 555-conjugated anti-Rat Ig antibody (1:200, Thermo Fisher Scientific, Waltham, MA, USA) in the dark for 1 h. After washing with PBS-T, the slides were mounted in ProLong antifade mounting solution with 4 ,6-diamidino-2-phenylindole dye (DAPI, Invitrogen, Waltham, MA, USA) and sealed with nail polish. The fluorescence signals of sfGFP (green), iNOS (red), and DAPI (blue) were detected using an LSM510 fluorescence microscope (Zeiss, Jena, Germany), and processed using LSM image software.

Statistical Analysis
Statistical analysis was done with Student's t-tests. A p-value < 0.05 was considered statistically significant. All data are expressed as mean ± SEM.

Study Design and Construction of Nitric Oxide (NO)-Sensing Gene Switch System
The NO-sensing gene switch system was designed as illustrated ( Figure 1). Briefly, the elevated NO generated by the increase in inducible nitric oxide synthase (iNOS) after bacterial colonization was used to trigger gene expression from the NO-sensing switch system in SL. ∆ppGpp (an attenuated S. Typhimurium defective in ppGpp biosynthesis). The switch system was constructed such that when fimS faced IRR, no gene was expressed, defined as the OFF state. When NO triggered the expression of FimE, leading to the reorientation of fimS, the GOI was expressed, defined as the ON-state.

Statistical Analysis
Statistical analysis was done with Student's t-tests. A p-value < 0.05 was considered statistically significant. All data are expressed as mean ± SEM.

Study Design and Construction of Nitric Oxide (NO)-Sensing Gene Switch System
The NO-sensing gene switch system was designed as illustrated ( Figure 1). Briefly, the elevated NO generated by the increase in inducible nitric oxide synthase (iNOS) after bacterial colonization was used to trigger gene expression from the NO-sensing switch system in SL. ΔppGpp (an attenuated S. Typhimurium defective in ppGpp biosynthesis). The switch system was constructed such that when fimS faced IRR, no gene was expressed, defined as the OFF state. When NO triggered the expression of FimE, leading to the reorientation of fimS, the GOI was expressed, defined as the ON-state. The NO-sensing gene switch system was constructed with two plasmids (Figure 2). One plasmid, with a high-copy pBR322 replication origin (ori), contained a fim switch block (IRR-fimS-IRL-GOI), with either Renilla luciferase variant 8 (rluc8) [36,37] or superfolder green fluorescent protein (sfGFP) [38] acting as the reporter gene, with strong ribosome binding site (RBS). The fimS gene containing a promoter region was located upstream of GOI in the reverse direction, with GOI expression triggered after the reorientation of fimS. This plasmid also contained the norR gene with a constitutive PlacIq promoter and the same RBS. This combination has been used to suppress additional copies of PnorV on exogenous plasmids [33]. The resulting plasmids containing rluc8 and sfGFP were named pSRluc8 and pSGFP, respectively. The NO-sensing gene switch system was constructed with two plasmids (Figure 2). One plasmid, with a high-copy pBR322 replication origin (ori), contained a fim switch block (IRR-fimS-IRL-GOI), with either Renilla luciferase variant 8 (rluc8) [36,37] or superfolder green fluorescent protein (sfGFP) [38] acting as the reporter gene, with strong ribosome binding site (RBS). The fimS gene containing a promoter region was located upstream of GOI in the reverse direction, with GOI expression triggered after the reorientation of fimS. This plasmid also contained the norR gene with a constitutive P lacIq promoter and the same RBS. This combination has been used to suppress additional copies of P norV on exogenous plasmids [33]. The resulting plasmids containing rluc8 and sfGFP were named pSRluc8 and pSGFP, respectively.
Another plasmid, pFimE, was responsible for the NO-driven expression of FimE recombinase. Leaky expression of FimE recombinase has often led to gene switching, even under non-sensing conditions [39], indicating that a small amount of FimE was sufficient to drive the reorientation of the fimS gene fragment. In the pFimE plasmid, the fimE gene without a typical RBS was under the control of the PnorV promoter for low level of expression and was inserted into pTU2S-a with a low-copy p15A ori [40].

Characterization of Bacteria Transformed with the NO-Sensing Switch System
To characterize the NO-responsive properties of the constructed switch system, diethylenetriamine/nitric oxide (DETA/NO) was used as an in vitro NO source.
NO levels released by DEAT/NO were measured in bacterial culture media after incubation for 16 h at 37 °C ( Figure 3A). NO levels ranged from ~3 μM in the presence of 25 μM DEAT/NO to ~12 μM in the presence of 1000 μM DEAT/NO.
The NO sensitivity of the switch system was evaluated by incubating SL. pNASR with various concentrations (0-1000 μM) of DEAT/NO, with same treated SL. pSRluc8 being the control ( Figure 3B-E). Switch activation was indicated by Rluc8 expression and Rluc8 intensity was measured ( Figure 3B,C). Bioluminescence signal was significantly higher at 50 μM DEAT/NO, increased dose-dependently, and reached a maximum at 200 μM DEAT/NO. By contrast, SL. pRluc8 transformed with pRluc8 alone did not show any signal, even at 1000 μM DEAT/NO. These results were consistent with that of western blot analysis of Rluc8 expression ( Figure 3D). The switch of fimS was also detected by PCR analysis using an ON-state primer set ( Figure 3E). No band was detected in SL. pSRluc8, even at 1000 μM DEAT/NO. In SL. pNASR, negligible bands were detected at no or low DETA/NO concentrations, which were likely due to the leaky expression of FimE and NO produced by bacteria themselves. However, the band intensities were markedly increased at 25 μM and higher DETA/NO concentrations. These findings indicated that activation of the NO-sensing switch system was dependent on elevated NO, that the switch was activated at 25 μM DETA/NO, and that it reached a detectable level at 50 μM. Another plasmid, pFimE, was responsible for the NO-driven expression of FimE recombinase. Leaky expression of FimE recombinase has often led to gene switching, even under non-sensing conditions [39], indicating that a small amount of FimE was sufficient to drive the reorientation of the fimS gene fragment. In the pFimE plasmid, the fimE gene without a typical RBS was under the control of the P norV promoter for low level of expression and was inserted into pTU2S-a with a low-copy p15A ori [40].

Characterization of Bacteria Transformed with the NO-Sensing Switch System
To characterize the NO-responsive properties of the constructed switch system, diethylenetriamine/nitric oxide (DETA/NO) was used as an in vitro NO source.
NO levels released by DEAT/NO were measured in bacterial culture media after incubation for 16 h at 37 • C ( Figure 3A). NO levels ranged from~3 µM in the presence of 25 µM DEAT/NO to~12 µM in the presence of 1000 µM DEAT/NO.
The NO sensitivity of the switch system was evaluated by incubating SL. pNASR with various concentrations (0-1000 µM) of DEAT/NO, with same treated SL. pSRluc8 being the control (Figure 3B-E). Switch activation was indicated by Rluc8 expression and Rluc8 intensity was measured ( Figure 3B,C). Bioluminescence signal was significantly higher at 50 µM DEAT/NO, increased dose-dependently, and reached a maximum at 200 µM DEAT/NO. By contrast, SL. pRluc8 transformed with pRluc8 alone did not show any signal, even at 1000 µM DEAT/NO. These results were consistent with that of western blot analysis of Rluc8 expression ( Figure 3D). The switch of fimS was also detected by PCR analysis using an ON-state primer set ( Figure 3E). No band was detected in SL. pSRluc8, even at 1000 µM DEAT/NO. In SL. pNASR, negligible bands were detected at no or low DETA/NO concentrations, which were likely due to the leaky expression of FimE and NO produced by bacteria themselves. However, the band intensities were markedly increased at 25 µM and higher DETA/NO concentrations. These findings indicated that activation of the NO-sensing switch system was dependent on elevated NO, that the switch was activated at 25 µM DETA/NO, and that it reached a detectable level at 50 µM.  The percentage of switch activated at various concentrations of DETA/NO was investigated using SL. pNASGFP carrying a sfGFP gene. (Figure S1). The percentages of fluorescent colonies were dependent on DETA/NO concentrations, with 20%, 70%, and 100% fluorescent colonies detected at DETA/NO concentrations of 100, 200, and 1000 μM, respectively. These results suggested that the phenotype of this NO-sensing switch system can be detected at 100 μM DETA/NO (~5 μM NO).
The timing of the gene switching event in the presence of DETA/NO was determined by incubating SL. pNASR with 100 μM DETA/NO and measuring Rluc8 intensity by PCR analysis ( Figure S2). In the presence of DETA/NO, strong Rluc8 intensity was detected in SL. pNASR after incubating for 30 min, consistent with the band detected by PCR test. These results indicated that NO induction of the switch event was very rapid.
FimE catalyzes the unidirectional switch of fimS, as it shows different binding affinities towards IRL and IRR, and is only able to invert fimS when it faces IRR. Thus, the switch inversion induced by FimE is both inheritable and permanent [27,33]. To determine the switch inheritance and permanency of the NO-sensing system, SL. pNASR was incubated overnight with 100 μM DETA/NO, followed by culture for an additional 72 h in the absence of DETA/NO (Figure 4). Rluc8 intensity was measured at the indicated time points. In the absence of DETA/NO pretreatment, the bioluminescence intensity in SL. pNASR was similar to that of control SL. pSRluc8 (0 h). DETA/NO pretreatment increased the Rluc8 intensity in SL. pNASR at 0 h by about 200-fold compared with other groups, The percentage of switch activated at various concentrations of DETA/NO was investigated using SL. pNASGFP carrying a sfGFP gene. (Figure S1). The percentages of fluorescent colonies were dependent on DETA/NO concentrations, with 20%, 70%, and 100% fluorescent colonies detected at DETA/NO concentrations of 100, 200, and 1000 µM, respectively. These results suggested that the phenotype of this NO-sensing switch system can be detected at 100 µM DETA/NO (~5 µM NO).
The timing of the gene switching event in the presence of DETA/NO was determined by incubating SL. pNASR with 100 µM DETA/NO and measuring Rluc8 intensity by PCR analysis ( Figure S2). In the presence of DETA/NO, strong Rluc8 intensity was detected in SL. pNASR after incubating for 30 min, consistent with the band detected by PCR test. These results indicated that NO induction of the switch event was very rapid.
FimE catalyzes the unidirectional switch of fimS, as it shows different binding affinities towards IRL and IRR, and is only able to invert fimS when it faces IRR. Thus, the switch inversion induced by FimE is both inheritable and permanent [27,33]. To determine the switch inheritance and permanency of the NO-sensing system, SL. pNASR was incubated overnight with 100 µM DETA/NO, followed by culture for an additional 72 h in the absence of DETA/NO (Figure 4). Rluc8 intensity was measured at the indicated time points. In the absence of DETA/NO pretreatment, the bioluminescence intensity in SL. pNASR was similar to that of control SL. pSRluc8 (0 h). DETA/NO pretreatment increased the Rluc8 intensity in SL. pNASR at 0 h by about 200-fold compared with other groups, indicating that the switch had been activated. Rluc8 intensity gradually increased, being more than 10-fold higher at 48 h than at 0 h but was lower at 72 h, which was thought to be due to bacterial cells being in the death phase of the growth curve. This result indicated that the switched gene was maintained permanently in the ON-state during bacterial division and plasmid replication, even in the absence of further NO stimulation after activation.
iosensors 2022, 12, x FOR PEER REVIEW indicating that the switch had been activated. Rluc8 intensity gradually more than 10-fold higher at 48 h than at 0 h but was lower at 72 h, which be due to bacterial cells being in the death phase of the growth curve. This that the switched gene was maintained permanently in the ON-state dur vision and plasmid replication, even in the absence of further NO stimu vation.

In Vivo Tumor-Specific Gene Delivery by Bacteria with NO-Sensing Gene
To assess whether SL. ΔppGpp bearing the NO-sensing gene switch cally colonized and induced immune response in tumors in vivo (Figur was intravenously injected into CT26 tumor-bearing mice. The numbers o terial cells on day 3 were about 10,000-fold higher in tumors (~10 9 colo (CFU)/g) than in normal organs, such as the liver, spleen, and lungs (~10 5A). The numbers of bacterial cells in tumors gradually decreased over tim higher in tumors than in other organs until day 11. The immune respo bacterial cells colonizing in tumors were evaluated by measuring the lev flammatory cytokines, including IFN-γ, TNF-α, and IL-1β, in the tumo 5B-D). The levels of TNF-α and IL-1β were significantly higher, beginnin the levels of IFN-γ were significantly higher, beginning on day 3, in bacte than in the PBS group, similar to previous reports [11,12].
The expression of iNOS can be upregulated in response to bacteria flammatory cytokines [15,20]. This was confirmed by measuring iNOS e in tumors treated with SL. pNASR ( Figure 5E,F). iNOS protein in tumors expressed on day 1, maximized on day 3 with up to a 500-fold increase, a day 7. These results confirmed that iNOS, an enzyme generating high

In Vivo Tumor-Specific Gene Delivery by Bacteria with NO-Sensing Gene Switch System
To assess whether SL. ∆ppGpp bearing the NO-sensing gene switch system specifically colonized and induced immune response in tumors in vivo ( Figure 5), SL. pNASR was intravenously injected into CT26 tumor-bearing mice. The numbers of colonizing bacterial cells on day 3 were about 10,000-fold higher in tumors (~10 9 colony forming unit (CFU)/g) than in normal organs, such as the liver, spleen, and lungs (~10 5 CFU/g) ( Figure 5A). The numbers of bacterial cells in tumors gradually decreased over time, but remained higher in tumors than in other organs until day 11. The immune responses induced by bacterial cells colonizing in tumors were evaluated by measuring the levels of the proinflammatory cytokines, including IFN-γ, TNF-α, and IL-1β, in the tumor tissues ( Figure 5B-D). The levels of TNF-α and IL-1β were significantly higher, beginning on day 1, and the levels of IFN-γ were significantly higher, beginning on day 3, in bacteria-treated mice than in the PBS group, similar to previous reports [11,12].
The expression of iNOS can be upregulated in response to bacterial LPS and proinflammatory cytokines [15,20]. This was confirmed by measuring iNOS expression levels in tumors treated with SL. pNASR ( Figure 5E,F). iNOS protein in tumors was found to be expressed on day 1, maximized on day 3 with up to a 500-fold increase, and decreased on day 7. These results confirmed that iNOS, an enzyme generating high amounts of NO, was highly and rapidly upregulated after bacterial colonization of tumors. The potency of this NO-sensing switch system for in vivo tumor-targ ery was evaluated ( Figure 6). CT26 tumor-bearing mice were intravenous 1 × 10 7 CFUs of SL. pNASR or SL. pSRluc8 as control. As expected, tumors showed no detectable levels of bioluminescence signal, as there was no N expression ( Figure 6A). By contrast, signals were detected in the tumor The potency of this NO-sensing switch system for in vivo tumor-targeted gene delivery was evaluated ( Figure 6). CT26 tumor-bearing mice were intravenously injected with 1 × 10 7 CFUs of SL. pNASR or SL. pSRluc8 as control. As expected, tumors in control mice showed no detectable levels of bioluminescence signal, as there was no NO-driven FimE expression ( Figure 6A). By contrast, signals were detected in the tumor region on day 3 after SL. pNASR injection, with these levels being increased, reaching a maximal level on day 7 ( Figure 6A,B). Despite iNOS expression being decreased on days 5 and 7, the Rluc8 intensity increased. These results also confirmed that once the switch system is activated, gene expression is maintained, even after changes in NO generation in the tumor microenvironment. This in vivo result was also consistent with the ex vivo Rluc8 signal in organs or tumor tissues obtained from mice treated with SL. pNASR ( Figure 6C,D). Beginning 3 days after injection, strong Rluc8 signals were detected in the tumor region, with high tumor-to-normal organ ratios, further confirming that genes delivered by this NO-sensing switch system are tumor-specific. To assess whether the NO-sensing gene switch system was activated by NO generated by iNOS expression, CT-26 tumor-bearing mice were intraperitoneally treated with the iNOS inhibitor 1400 W dihydrochloride once daily [41], beginning 3 days before SL. pNASR injection until the end of the experiment, and bioluminescence images of the mice were taken after bacterial injection on the indicated days (Figure 7). The signals in the To assess whether the NO-sensing gene switch system was activated by NO generated by iNOS expression, CT-26 tumor-bearing mice were intraperitoneally treated with the iNOS inhibitor 1400 W dihydrochloride once daily [41], beginning 3 days before SL. pNASR injection until the end of the experiment, and bioluminescence images of the mice were taken after bacterial injection on the indicated days (Figure 7). The signals in the tumor tissues were lower in mice than that were not treated with 1400 W on day 3, with the difference being significant on days 5, 7, and 9 ( Figure 7B,C). This finding indicated that this switch system was specifically induced by NO produced by iNOS, a finding confirmed by co-localization of iNOS and GOI expression via immunofluorescence staining. SL. pNASGFP was used instead of SL. pNASR to visualize gene expression in tumors, with switch activation indicated by the expression of sfGFP. Strong sfGFP signals were detected at sites with increased iNOS expression from tumor sections on days 2, 3, and 5 ( Figure 7D). These results indicated that the gene switch system described in this study could be specifically activated by NO in tumors after bacterial colonization, and can realize tumortargeted gene delivery. Figure 7. In vivo NO−specific switch activation. (A) Scheme of the experiment. CT26 tumorbearing mice were intraperitoneally administered 1400 W daily, starting 3 days before SL. pNASR injection until the end of the experiment. Bioluminescence imaging was performed 3, 5, 7, and 9 days after administration of SL. pNASR. (B) Bioluminescence images of CT26 tumor-bearing mice in the presence or absence of 1400 W treatment 3, 5, 7, and 9 days after injection of 1 × 10 7 CFUs SL. pNASR. (C) Quantification of bioluminescence signals in the tumor tissues. Values represent the mean ± SEM. * p < 0.05, ** p < 0.01, and *** p < 0.001; ns, not significant by Student's t-tests. (D) Immunofluorescence staining of iNOS and sfGFP expression in CT-26 tumor tissues on the indicated days after SL. pNASGFP injection. Switch activation was indicated by the expression of sfGFP (green), sections were stained with anti-iNOS antibody (red), and nuclei were stained with DAPI (blue). Merged images are also shown (merged). Scale bar, 50 µm. Data are representative of three independent experiments.

Discussion
These results indicated that the gene switch system described in this study could be specifically activated by NO in tumors after bacterial colonization, and can realize tumor-targeted gene delivery.

Discussion
The present study used SL. ∆ppGpp with a nitric oxide (NO)-sensing switch system to show that elevated NO in tumors generated by upregulated inducible nitric oxide synthase (iNOS) after S. Typhimurium infection can induce tumor-targeted gene delivery. Sensing of NO by the switch system led to the expression of a site-specific DNA recombinase, which induced the permanent inversion of a promoter region, resulting in the sustainable expression of a target gene.
Although tumor-targeting bacteria can reach high tumor/normal organ ratio, the presence of bacterial cells in normal tissue upon systemic administration could damage normal tissues. To reduce non-specific gene expression and off-target effects, efforts have been made to regulate payload expression by placing the gene under the control of a promoter responding to a specific signal given at proper time points, rather than using constitutively promoters [7,42]. These signals can be external chemicals, such as L-arabinose to induce the P BAD promoter [43] and tetracycline that triggers a bidirectional P Tet promoter [44]. Multiple doses of these inducers are required to induce and maintain the continuous expression of the payloads, which have been found to cause side effects in mice [45]. Endogenous signals from unique tumor microenvironments have also been developed, including hypoxia [46] and acidic pH in tumors [47]. Endogenous inducers may overcome the drawbacks associated with exogenous inducers, but changes in the tumor microenvironment during treatment may also affect therapeutic efficacy.
Developments in synthetic biology have enabled the design of more complex gene circuits [48]. A memory circuit, which was designed to record cellular events via permanent changes to DNA caused by recombinase in response to a transient input signal, was used to engineer diagnostics and controllable sustained therapeutic expression [49]. Use of a recombinase-based memory circuit can convert a temporary trigger to a long-term cellular response, overcoming the requirement for continuous stimulation of the inducible promoters.
The present study describes the construction of a NO-activated memory circuit, resulting in long-term NO-driven gene expression carried by SL. ∆ppGpp. NorR detection of NO was found to lead to the expression of a DNA recombinase, FimE, which catalyzed the permanent and inheritable inversion of a DNA switch, fimS, leading to the sustainable expression of the GOI. This switch system was shown to respond to chemical source of NO (DETA/NO, diethylenetriamine/nitric oxide) in a rapid and dose-dependent manner in vitro. Removal of DETA/NO after switch activation did not affect Rluc8 expression in bacterial cells, indicating that the activated switch can be maintained and passaged via replication. In vivo NO-induced tumor target gene delivery was tested by bioluminescence imaging. This switch could be specifically activated in tumors after bacterial colonization, with a high tumor-to-normal organ ratio, and 1400 W inhibition of iNOS activity demonstrated that switch activation was specific to NO generated by iNOS. These results confirmed that NO can be used as an inducer for tumor-specific gene delivery by S. Typhimurium.

Conclusions
Taken together, our study showed that the increased NO in tumors after SL. ∆ppGpp colonization can induce tumor-specific gene delivery. The permanent and inheritable switch activation caused by DNA recombinase enabled sustained gene expression in tumors. This NO-sensing switch system could represent a promising strategy for tumor-targeted gene delivery for diagnostic and/or therapeutic purposes.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/bios13020266/s1, Figure S1: Switch activation percentage under various concentrations of DETA/NO. Figure S2: Time-dependent switch event in bacterial cells with the NO−sensing switch system. Table S1: Plasmids and SL. ∆ppGpp transformants.