| BNUZH-China - iGEM 2023

Overview

Our project aims to provide a broad-spectrum cancer treatment approach. The main focus of our laboratory work was to demonstrate the effectiveness of our bacteria-based ferroptosis-inducing cancer therapy. Specifically, we engineered Salmonella typhimurium VNP20009 to promote Fenton reaction and the silencing of SLC7A11 gene, which could cause ferroptosis in cancer cells. To enhance the specificity of the engineered bacteria to tumor cells, we designed a targeting module based on the single chain variable fragment. To ensure the safety of our project’s design, we designed a gene circuit (a toxin-antitoxin system) to prevent plasmid loss and to manipulate the initiation of suicide of the engineered bacteria. The combination of experiment and modeling further demonstrated the feasibility and effectiveness of our design.

Technical Feasibility

Generation of H₂O₂ to promote Fenton reaction

We engineered bacteria VNP20009 to express the glucose oxidase (GOx). Our experimental results demonstrated the successful expression and delivery of GOx into tumor cells by type III secretion system (T3SS). Tumor cells were infected with VNP20009 and VNP-SopE-GOx for 24 hours. Culture supernatant was then collected for evaluating the H₂O₂ content. The experiment clearly demonstrated that the H₂O₂ level in the culture medium supernatant of tumor cells infected by VNP-SopE-GOx was significantly higher compared to that of the control group and that of tumor cells infected by VNP20009 (Figure 1). Previous studies have indicated that hydrogen peroxide produced in cells can be transported to the outside of cells through aquaporins (AQPs), resulting in increased hydrogen peroxide content in the microenvironment¹. We further validated the significant induction of ferroptosis in tumor cells through VNP-SopE-GOx infection utilizing the cell counting kit-8 assay (Figure 2).

To see the full result, please turn to the RESULT page.

Figure 1. H₂O₂ levels in the cell culture supernatant

Figure 2. Cell viability of cells treated with bacterial infection and co-treated with bacterial infection and Fer-1.

Delivery of shRNA-SLC7A11 into tumor cells by bacteria infection

To validate the feasibility of delivery of shRNA into tumor cells by bacteria infection, we cultured human gastric adenocarcinoma cells BGC-823 and transiently transfected with the eukaryotic GFP-expressing vector. Compared with the VNP20009 control, the fluorescence intensity of tumor cells was significantly weakened after VNP20009-shGFP infection. The results indicated that the engineered bacteria successfully delivered shRNA-GFP into tumor cells and effectively downregulated the expression of GFP (Figure 3). Then, we transfected BGC-823 cells with VNP20009-shSLC7A11 to induce SLC7A11 gene silencing. Meanwhile, we also used siRNA transfection with BGC-823 as a positive control. qRT-PCR results showed that SLC7A11 gene expression was significantly down-regulated in tumor cells compared with control cells (Figure 4). In conclusion, it was feasible to use engineered bacteria for gene silencing by delivering shRNA, and the system used in in our project worked efficiently.

To see the full result, please turn to the RESULT page.

Figure 3. The fluorescence intensity of tumor cells infected by bacteria for 24 hours. Both the engineered bacteria and VNP20009 infected tumor cells at the MOI of 1:500. Control was BGC-823 not transfected with the GFP plasmid.

Figure 4. The expression levels of SLC7A11. a. The expression level of SLC7A11 of BGC-823 after siRNA transfection. Control was BGC-823 without any treatment. siRNA-control was siRNA targeting another gene. Results were significant between the siRNA-SLC7A11 and control groups (*** p < 0.001). b. The expression levels of SLC7A11 of BGC-823 after infected by engineered bacteria at different MOI. VNP20009 without functional plasmids at a MOI of 1:5000 was used as control group. The expression of SLC7A11 between VNP-shSLC7A11 (1:3000), VNP-shSLC7A11 (1:5000), and VNP-control group showed significant differences (* p < 0.05, ** p < 0.01)

Expression of anti-CEA scFv in VNP20009

To enhance the targeting specificity of our engineered bacteria, we planned to use the single chain fragment variable (scFv) against carcinoembryonic antigen (CEA). A GFP tag was added to the C-terminus of Lpp-OmpA-scFv on the plasmid, in order to characterize the expression of anti-CEA scFv. Western blotting proved that Lpp-OmpA-scFv-GFP fusion protein could be expressed in VNP20009 (Figure 5).

Co-culture of the engineered bacteria and tumor cells was performed to verify the efficiency of anti-CEA scFv in guiding the engineered bacteria to infect CEA high expressing tumor cells. Finally，we chose human colon cancer cells line LS174T with high-CEA-expression as experimental group, and a human gastric cancer cell line BGC-823 with low-CEA-expression as CEA negative cell lines. The engineered bacteria with GFP tag and the negative control with RFP tag were used to infect the above two types of cells, and the function of anti-CEA scFv was verified by the infection efficiency of the bacteria.Because LS174T cells were prone to stack growth during culture, and with the extension of infection time, the morphology of LS174T cells changed to a certain extent. However, by analyzing the number of the engineered bacteria with two colors of fluorescence by fluorescence microscopy, we found that the engineered bacteria expressing scFv could target the LS174T cells with high CEA expression better than BGC-823 cells, while the strains without scFv expression had no obvious selection bias (Figure 6).

To see the full result, please turn to the RESULT page.

Figure 5. WB analysis of the expression of specific single chain antibody fragments (scFv). Lpp-OmpA-scFv-GFP molecular weight is about 66 kDa. The upper band shows the expression of Lpp-OmpA-scFv-GFP fusion protein in VNP20009.

Figure 6. Microphotographs of BGC-823 and LS174T cells co-infected by engineered bacteria and VNP2009 for 2 hours. Both the engineered bacteria and negative control infections were at the MOI of 1:50.

The toxin-antitoxin system

To validate the “logical suicide circuit”, we first tested the functionality of the pLacO and pLtetO promoters.Aiming to test the functionality of the pLacO and pLtetO promoters, we constructed the pET-GFP-LacI and pJKR-L-tetR plasmids. The pET-GFP-LacI plasmid utilized the constitutive promoter prpsM to express the LacI protein, which effectively represses the pLacO promoter. Meanwhile, pLacO regulates the expression of GFP. Similarly, the pJKR-L-tetR plasmid contains a tetR repressor protein that efficiently represses the pLtetO promoter, with pLacO governing the expression of GFP. We separately transformed the pET-GFP-LacI and pJKR-L-tetR plasmids into BL21 (DE3) cells to validate a portion of our logical circuit (called BL21-pET-GFP-LacI and BL21-pJKR-L-GFP-tetR). By measuring the fluorescence intensity of the cultured cells, we verified that pLacO and pLtetO could effectively work (Figure 7a). We also found that the optimal concentration of doxycycline was about 0.1 μg/mL (Figure 7b). The test of the functionality of pET-Hok-Sok-lacI plasmid showed that the engineered bacteria could not grow normally on LB-agar plates without IPTG while could grow normally on LB-agar plates with IPTG (Figure 8). The above results preliminarily proved the feasibility of the "logical suicide circuit" .

Figure 7. a. BL21-pET-GFP-lacI was BL21 (DE3) with pET-GFP-lacI plasmid and control was original BL21 (DE3). The fluorescence intensity of BL21-pET-GFP-LacI+IPTG was significantly higher than that of BL21-pET-GFP-LacI and control. b, BL21-pJKR-L-GFP-tetR was BL21 (DE3) with pJKR-L-tetR plasmid and control was original BL21 (DE3). The fluorescence intensity of BL21-pJKR-L-GFP-tetR+Dox are significantly higher than BL21-pJKR-L-GFP-tetR and control especially BL21-pJKR-L-GFP-tetR+0.1 μg/mL Dox.

Figure 8. Bacterial BL21-Hok/Sok can grow in the LB-agar plates with IPTG(left), but cannot live in the LB-agar plates without IPTG(right).

To have a whole comprehension of our wet lab work, please turn to EXPERIMENTS and RESULTS pages.

Model Prediction

The toxin-antitoxin system

Through modeling, we made further predictions about the effectiveness of the toxin-antitoxin system in practical applications.

We developed Gene Circuit Model to verify the functions of the toxin-antitoxin system in preventing plasmid loss and doxycycline inducing the engineered bacteria to suicide. By utilizing ordinary differential equations to simulate the functional processes, the model predicted the concentration of toxin Hok in the case of the plasmid having been lost or doxycycline having been introduced.

The results of Gene Circuit Model showed that after the engineered bacteria lost the plasmids (pFenton&pSilence) or after doxycycline was added, the reactions in the logical suicide circuit would lead to a high level of toxin Hok, exceeding the prespecified threshold, which was considered as the engineered bacterial suicide (Figure 9~11).

Figure 9. The concentration of Hok (after the introduction of doxycycline)

Figure 10. The concentration of Hok (after losing the plasmid pSilence)

Figure 11. The concentration of Hok (after losing the plasmid pFenton)

These results suggested that the toxin-antitoxin system could prevent plasmid loss and control the suicide of engineered bacteria by doxycycline.

Meanwhile, Plasmid Loss Model was established as a complement to Gene Circuit model to further demonstrated the maintenance plasmid function of the toxin-antitoxin system from practical effects. The passage process of engineering bacteria was simulated by using the recursive formula to predict the proportion of the engineered bacteria that maintained the plasmids after a period of time.

According to the results of Plasmid Loss Model, the proportion of the engineered bacteria that maintained the plasmids still maintained a high level in the presence of toxin-antitoxin system while that of in the case of without toxin-antitoxin system dropped rapidly (Figure 12). We also tried to adjust the killing probability of the toxin-antitoxin system against the bacteria that have lost the plasmid (parameter k) and to observe the sensitivity of the model to parameter k. The results showed that the plasmid loss rate was significantly reduced after the introduction of the toxin-antitoxin system at multiple k values ranging from 0.92 to 0.98 (Figure 13).

Figure 12. Plasmid Maintenance Rates

Figure 13. Plasmid Maintenance Rates under Multiple k Values (0~100 h amplification)

The results suggested that within a certain margin of error, the maintenance of plasmid function in our toxin-antitoxin system was very efficient. Thus, our toxin-antitoxin system could effectively maintain plasmid.

The above results of model further illustrated the feasibility of the toxin-antitoxin system.

Validation of single chain antibody fragment (anti-CEA scFv)

We conducted molecular modeling analysis on the single-chain antibody fragment variable (scFv) employed in this project to investigate the interaction between the Lpp-OmpA-scFv and the carcinoembryonic antigen (CEA) present on the surface of tumor cells.

Since the structure of the Lpp-OmpA-scFv is unknown, our first step is to predict the structure of the antibody using AlphaFold2 (AF2). Upon analyzing the prediction results, we downloaded the optimal prediction (Figure 14.A). The AlphaFold prediction provides expected position error values for each residue pair (X.Y), showing the predicted position error at residue X when aligned with residue Y in the true structure. These residue-residue "predicted aligned error" values can be visualized with an error plot (Figure 14.B)

Figure 14. Prediction of Lpp-OmpA-scFv protein structure based on Alphafold2. (A)Best structural prediction of proteins of Lpp-OmpA-scFv

Figure 14. Prediction of Lpp-OmpA-scFv protein structure based on Alphafold2. (B)Predicted aligned error plot

To predict the transmembrane protein regions of the above structures and verify that anti-CEA scFv could be anchored to the engineered bacterial outer membrane, protein model analysis was performed. Additionally, we modified the membrane properties by utilizing the Analyze Transmembrane Proteins tools in Discovery Studio. We incorporated an implicit membrane into the protein structure. The Hidden Markov Model (HMM) was employed to predict the transmembrane helices based on the amino acid sequence of the protein. Subsequently, a hidden membrane consisting of two parallel planes was introduced to the protein structure (Figure 15). The placement of the membrane was determined by optimizing the simplified solvation energy. If there was a significant charge difference between protein residues located outside the membrane, adjustments were made to the membrane.

Figure 15. Transmembrane protein regions prediction of Lpp-OmpA-scFv

Notes: The angle of inclination is defined as the angle between the first principal axis of the protein and the normal of the membrane, and the angle of rotation is the angle between the second principal axis of the atomic set and the normal of the plane defined by the first principal axis and the normal of the membrane.

In order to verify that the anti-CEA scFv secreted by the engineered bacteria could bind to the high expressed carcinoembryonic antigen on the surface of tumor cells with a strong affinity, we employed molecular docking and corresponding calculations to predict the structure of the antibody and its affinity for CEACAM5. ZDOCK 3.0.2 was employed to dock CEACAM5 and Lpp-OmpA-scFv proteins, and the most optimal docking result was selected upon completion. The PyMol V2.4.0 software was utilized to label and present the binding sites of the docking complex (Figure 16 A). We particularly emphasized amino acids capable of forming hydrogen bonds on the interaction surface (Figure 16 B). Following that, we uploaded the docking complexes to PDBePISA in order to analyze the interaction domains and surfaces of the proteins involved (Table 1).

Figure 16. The docking of antibody molecules with antigen molecules. (A).Results of molecular docking, Lpp-OmpA-scFv is shown in green and CEACAM5 is shown in orange

Figure 16. The docking of antibody molecules with antigen molecules. (B)Schematic representation of the molecular docking surface. Among them, amino acids that can form hydrogen bonds are shown

Table 1. Summary of docking complex interface

To see the full result of our model, please turn to the MODEL page.

Proof Of Concept