In order to heal the intestinal tract damaged by IBD, we adopted a special therapy expressing the therapeutic proteins by E.coli Nissle 1917 (EcN) in situ. The design is based on a ternary system: sensor - secretion peptide - therapeutic proteins. Thus, EcN can be induced by inflammatory signals to secrete therapeutic proteins in the damaged intestine, promoting the intestinal mucosal healing.
For each of the three sections, we can replace with different parts freely for a more comprehensive treatment system. Therefore, there are various of combinations, making the whole method more flexible, so that we can provide the treatment with multi-aspects.
Fig.1 General design of the treatment ternary system
Fig.2 Big picture of mucosal healing project
Therapeutic proteins part
The primary task of mucosal healing is to find therapeutic proteins.
We planned to use our engineered probiotics EcN for the in situ secretion of therapeutic proteins to promote the recovery of wounds. In this way, the therapeutic protein directly secreted by the colonized probiotics EcN can be highly effective in treatment, and the therapeutic effect can persist with the colonization of probiotics[1,2].
On the other hand, this therapeutic proteins secretion project can be of great dimensions and multiple aspects. Through literature research, a series of IBD candidate therapeutic proteins were screened out by us. We ensured that those candidates have therapeutic effects on IBD, and they all have been reported to be expressed in E.coli. The wide variety of therapeutic protein candidates improves the flexibility of the therapy, so that a combination of therapeutic proteins can be used in real scenarios to carry out a multidimensional and multi-aspects treatment.
Fig.3 In situ therapy for IBD with therapeutic protein
To sum up, therapeutic proteins secretion strategy gives an economical and efficient potential direction of treatment for IBD, which holds great promise for mankind to conquer IBD eventually. There are several proteins which may have a therapeutic effect on IBD.
Table 1 List of candidate therapeutic proteins
All kinds of the protein mentioned above have been reported to express in E.coli, the details of each protein are shown in the part pages.
With this information in mind, we moved forward to work on these candidate proteins. The workflow is shown in Fig.4.
Fig.4 Work-flow of the proof-of-concept in therapeutic protein
Firstly, before we can start, we need to do codon optimization, because different species have different codon preferences, and without proper optimization, expression efficiency will be greatly reduced. Codon preference analysis was performed for all candidate proteins listed in Table 1 (see Fig.5). All of these proteins are worth studying, but we only chose a few proteins as a proof of concept in our actual wet-lab experiments because of limited time and the high expense of gene synthesis. Considering the size of the secreted protein and the cost in gene synthesis, we finally took TFF as an example and carried out our further experiments.
Given that other teams may also come across the same codon preference researches, we also developed a practical software tool for this kind of analysis, see Software P2N for details.
Fig.5 Codon preference confidence analysis for therapeutic protein, in theory, the total GC% of EcN is 49.13%, 1st letter GC% is 55.38%, 2nd letter GC% is 42.34%, and 3rd letter GC% is 50.58%. We compare P2N and GenScript® online codon preference tool (GenSmart) analysis results for the bias from theoretical values. The lighter the squares are, the better for the codon optimization. (DNA sequence of each protein is detailed in the part page)
Secondly, molecular cloning of the 3 TFF genes (TFF1, TFF2 and TFF3) , which were synthesized by biological companies, was completed. We constructed the fusion proteins csgA-TFF1, csgA-TFF2 and csgA-TFF3 (csgA is a special kind of secretion peptide, which will be mentioned later) under the tac promoter. Western blot was performed to verify the successful expression of the fusion proteins.The result is shown in Fig.6 (GADPH as internal references).
Fig.6 Western blot for csgA-TFF2 and TFF3 (GADPH as internal references)
After verifying that TFF3 can be successfully expressed in EcN, we also carried out preliminary animal experiments to verify its therapeutic effect. As is shown in Fig.7, the CB57BL/6 mice were induced to be IBD model by feeding DSS (3% solved in water), a widely used IBD molding agent.
Then the mice are divided into 5 groups, and the normal saline (negative control group #1), EcN bacterial containing GFP gene (blank control group #2), EcN bacterial containing TFF3 gene (TFF3 non-treated experimental group #3), chitosan coated EcN (i.e. EcN@PCS, see Delivery for details
) bacterial containing TFF3 gene (TFF3 treated experimental group #4) and salazosulfapyridine (positive control group #5) were applied to mice by intragastric administration on day 1, 3 and 5. The fecal occult blood representing IBD severity were measured before and after intragastric administration as a contrast. Additionally, the colon length was measured on the last day as another index of IBD severity.
Fig.7 The process of animal therapeutic experiment. TFF3@PCS means the chitosan coated EcN, which is our delivery strategy.Click here to see the detailed design.
Inferring from the weight decrease (Fig.8) and the severe fecal occult blood index (see Table 2) the mice were induced to have IBD around day 0. However, the weight of all 5 groups increased after that. This indicates a recovery from IBD and a disruption of molding. Such interruption affected the following therapeutic experiment and the weight variant of the control group and treatment group showed a similar tendency after each time of intragastric administration (Fig.8).
Fig.8 Observation of mice weight variance
Then we compared the fecal occult blood before and after applying saline, engineered bacterial or salazosulfapyridine (Table 2). The difference between the recovery of negative control and treatment groups are not significant. Only the positive control that were fed with salazosulfapyridine known to be capable to cure IBD shows a better recovery result.
Table 2 The fecal occult blood index before and after treatment
(Severity is classified as -, +, ++, +++ and ++++ (most severe), N/A represents a failure of feces collection)
As the IBD would also result in a shorter colon length, we also measured the colon length of mice to see the therapeutic effect (Fig.9). The difference between each group shows no significance due to the low amount of data and the incompletion of IBD molding, despite that the mean value of the treatment group seems a little higher than the negative control.
Fig.9 The measurement of mice colon length
Up to this point the in vivo experiment seems to be incapable to prove the therapeutic effect of TFF3. And we suppose that the incomplete IBD modeling causing the recovery of the mice may shade the treatment effect of the engineered bacterial. Such a problem will be fixed in future experiments with a larger sample size to obtain experimental data with more translational significance.
Secretion peptide part
One of the most essential issues is that our therapeutic protein must be secreted outside the engineered bacteria and diffuse inside the patient's intestinal tract. As we all know, bacteria like E. coli Nissle 1917 usually don’t have the ability to initiatively secrete engineered proteins. We need the fusion of a special kind of secretion peptides to help the therapeutic protein secrete.
Fig.10 Therapeutic proteins can cross the membrane and diffuse into gut with the help of secretion peptide
We have searched a huge series of secretion peptides and done codon optimization (see Table 3). As mentioned above, "csgA" of "csgA-TFF1/2/3" is such a kind of secretion peptide.
Table 3 Candidate signal peptide for E.coli
All kinds of the protein mentioned above have been reported to express in E.coli, the details of each protein are shown in the part pages.
The workflow for secretion peptide candidates is shown in Fig.11.
Fig.11 Work-flow of the proof-of-concept in secretion peptides.
Firstly, we have done codon optimizations using our software tool P2N designed for codon preference analysis (see Software P2N for details). The codon preference confidence analysis results are shown in Fig.12. A series of candidates were taken into analysis and the sequences were optimized as shown in Table 3. Then, all of the secretion peptides’ genes were chosen to be synthesized by biological companies.
Fig.12 Codon preference confidence analysis for secretion peptide, in theory, the total GC% of EcN is 49.13%, 1st letter GC% is 55.38%, 2nd letter GC% is 42.34%, and 3rd letter GC% is 50.58%. We compare P2N and GenScript® online codon preference tool (GenSmart) analysis results for the bias from theoretical values. The lighter the squares are, the better for the codon optimization. (DNA sequence of each protein is detailed on the part page)
Secondly, we constructed the plasmid. We first tested the functions of these secretion peptides by fusing them to GFP protein and checking the fluorescence signal from the peptide-GFP fusion protein. The bacteria supernatant was centrifuged at 8000 rpm for 1 minute. The fluorescence was then measured by the microplate reader, and the results are shown in Fig.13.
The secretion effiency of different signal peptide, the p-values above the bar are the t-test p-values compared to the RGP-GFP group. RGP-GFP group is the blank control group, and RGP-TFF3 is the negative control group. (RGP is a kind of plasmid backbone in our design)
Among these signal peptides, all kinds of the signal peptides are are worthy of further study and optimization. However, due to the time and cost limiting, and considering that csgA had the most obvious secretion effect, we finally chose csgA as our secretion peptide for further experiments. That's why we chose the csgA-TFF as our therapeutic fusion protein.
In E.coli, the protein with csgA can be transported to periplasmic space by Sec system. Then it will cross the outer membrane through the secretion channel (csgG) with the help of csgC and csgE (Fig.14).
Fig.14 The mechanism of csgA secretion.
The therapeutic protein secreted by the engineered bacteria needs to be strictly and delicately controlled to avoid any possible safety problems since the final working environment is the GI tract, and some therapeutic proteins (e.g. The trefoil factor) are carcinogenic and may cause safety problems. We also hope that the therapeutic protein can be secreted when it should be secreted to create spatiotemporal heterogeneity based on demand.
Therefore, a strict regulatory system is strongly needed. To achieve this, we want to create several sensor parts as an intelligent switch to conditionally regulate the expression of therapeutic proteins, ideally, in gut.
Fig.15 No inflammation signal in the healed gut, no expression of therapeutic protein
Is there something unsual in the gut or in the inflamed gut?
According to the literature research, we find that there are already some markers for gut inflammation diagnose. For example, thiosulfate and tetrathionate can be two appealing targets, for they are two important products in gut sulfur metabolism, which is reported to link to gut inflammation, while nitric oxide (NO) can be a general marker for inflammation.
As a result, people have designed serveral sensors based on the markers, thsS/R for sensing thiosulfate, ttrS/R for sensing tetrathionate and NorR for sensing nitric oxide.
Fig.16 Principle for thrS/R and ttsS/R sensing system. Figure reference from BBa_K2507004 and BBa_2507006
Surprisingly, these three sensor parts have already been constructed by other iGEM teams before:
Table 4 Three inflammation sensor parts created by iGEMers before
Fig.17 Sensor binds to inflammation signals and initiate therapeutic protein expression
Proof of concept
To test the function of these three sensor parts, we used GFP element as a marker. We got pSB4K5-thsS (BBa_K2507000) & pSB1C3-thsR-sfGFP (BBa_K2507001), pSB4K5-ttrS(BBa_K2507002) & pSB1C3-ttrR-sfGFP (BBa_K2507003) from Guoqiang Chen's lab and assembled PnorV (NorR promotor from Nissle 1917 genome) into RGP-EGFP plasmid and substituted lac-element while retained RiboJ element.
Fig.18 Plasmid map model for RGP-PnorV-riboJ-EGFP
Table 5 Basic information of the 3 sensors we used
All the parts were tried to transform into EcN. For those two double plasmid systems--thsS/R and ttrS/R, firstly we tried to do co-electrotransformion but failed for the lower frequency of co-infection or inconsistent stoichiometry.
Fig.19 Co-electrotranformation of pSB4K5-ths/ttrS and pSB1C3-ths/ttrR-sfGFP to Nissle 1917
Another strategy is to successively electrotransform pSB4K5-ths/ttrS and pSB1C3-ths/ttrR-sfGFP into the bacteria. For this strategy, preparation of competent cells after the first successful plasmid transformation is needed, which is so cumbersome that we only made some attempts and then moved to a more secure strategy
Fig. 20 Successive electrotranformation of pSB4K5-ths/ttrS and pSB1C3-ths/ttrR-sfGFP to Nissle 1917
We then tried to do the co-tranfromation in the strain BL21, and many colonies were successfully grown on the Kanamycin-Chloramphenicol plate, with correct sequencing result.
Characterization of the sensor parts were carried out by bacterial flow cytometry for the fluorescence value is not affected by the amount of cells.
Table 6. Group setting for sensor characterization
With RGP-EGFP (RGP is the plasmid backbone in our design) transformed BL21 strain as a positive control, unmodified RGP plasmid transformed BL21 as the negative control, the cells with flurosence were sorted and counted, and the results are shown in the figure below, where we can see thsS/R show the most cheerful result, with a low rate of leaking and a high level of inducer-activation that are consistent with the expectation.
Fig.21 Work-flow of the proof-of-concept in sensor
Fig.22 Flow cytometry verification of sensor parts
RiboJ, a self-splicing ribozyme, is a common used insulator. With a hairpin structure, it can reduce unexpected interactions between neighboring sequences in a genetic circuit. Some reports also say insulation with RiboJ can increase downstream gene transcript abundance.Thus, we intend to insert RiboJ behind the sensor part, in other words, construct Pths-RiboJ(BBa_K3924041), Pttr-RiboJ(BBa_K3924040), PnorV-RiboJ(BBa_K3924042) three composite parts, to improve the performance of our sensors.
To make our long story short, we have provided a completely new method to treat IBD by expressing the therapeutic proteins in situ from engineered bacteria. We believe this strategy is highly promising in IBD mucosal healing. The project is based on the ternary system we created: sensor - secretion peptide - therapeutic protein. As a part collection, this innovative system will provide a conceptually novel way for curing IBD.
We have made several important progress on this project, which highlights the potential of our therapeutic method in future IBD treatment.
Firstly, we have completed all of the codon preference analysis of therapeutic protein candidates and secretion peptide candidates. These optimizations are based on the codon preference system developed on our own. We have also shared this codon preference software tool in our wiki for the convenience of future iGEM teams carrying out related researches.
Secondly, the basic molecular cloning have been accomplished and the protein expression based on csgA-TFF3 from E.coli Nissle 1917 have been proved to be a success.
Thirdly, we have constructed various of secretion peptides into our reporter plasmid. The fluorescence we measured have verified that the csgA peptide we chose are practical.
Fourthly, 3 kinds of sensors have been taken into demonstration and the completion of our whole ternery system is basically realised to this extent.
Last but not least, we have already began our further verification on C57BL/6 with IBD model. The experiments are still on-going.
In the future, we plan to continue our investigation into therapeutic proteins other than TFF3, while using the same paradigm as in the research of TFF3. In addition, we decided to go on our demonstration experiment on C57BL/6 with DSS induced enteritis model to measure the therapeutic effect. Moreover, the optimization for IBD sensors will also be taken into consideration. Hopefully, we will develop an intact, promising and practical treatment system in the end that might potentially relief the pain of IBD patients.
Schultz, M., Watzl, S., Oelschlaeger, T. A., Rath, H. C., Göttl, C., Lehn, N., Schölmerich, J., & Linde, H. J. (2005). Green fluorescent protein for detection of the probiotic microorganism Escherichia coli strain Nissle 1917 (EcN) in vivo. Journal of microbiological methods, 61(3), 389–398. https://doi.org/10.1016/j.mimet.2005.01.007
Joeres-Nguyen-Xuan, T. H., Boehm, S. K., Joeres, L., Schulze, J., & Kruis, W. (2010). Survival of the probiotic Escherichia coli Nissle 1917 (EcN) in the gastrointestinal tract given in combination with oral mesalamine to healthy volunteers. Inflammatory bowel diseases, 16(2), 256–262. https://doi.org/10.1002/ibd.21042
Aamann, L., Vestergaard, E. M., & Grønbæk, H. (2014). Trefoil factors in inflammatory bowel disease. World journal of gastroenterology, 20(12), 3223–3230.
Praveschotinunt, P., Duraj-Thatte, A.M., Gelfat, I. et al. Engineered E. coli Nissle 1917 for the delivery of matrix-tethered therapeutic domains to the gut. Nat Commun 10, 5580 (2019).
La Manna, S., Di Natale, C., Florio, D., & Marasco, D. (2018). Peptides as Therapeutic Agents for Inflammatory-Related Diseases. International journal of molecular sciences, 19(9), 2714.
Li, M. C., & He, S. H. (2004). IL-10 and its related cytokines for treatment of inflammatory bowel disease. World journal of gastroenterology, 10(5), 620–625.
Shukla, P. K. , Meena, A. S. , Rao, V. , Rao, R. G. , Balazs, L. , & Rao, R. K. . (2018). Human defensin-5 blocks ethanol and colitis-induced dysbiosis, tight junction disruption and inflammation in mouse intestine. entific Reports, 8(1).
Duan, L., Rao, X., Braunstein, Z., Toomey, A. C., & Zhong, J. (2017). Role of Incretin Axis in Inflammatory Bowel Disease. Frontiers in immunology, 8, 1734.
 Zhu, Y. , Jie, Z. , Yi, F. , Chen, L. , Zhang, L. , & Fei, Y. , et al. (2018). Control of intestinal inflammation, colitis-associated tumorigenesis, and macrophage polarization by fibrinogen-like protein 2. Frontiers in Immunology, 9, 87-.
Guidi, L., Mocci, G., Marzo, M., & Rutella, S. (2008). Treatment of Crohn's disease with colony-stimulating factors: An overview. Therapeutics and clinical risk management, 4(5), 927–934. https://doi.org/10.2147/tcrm.s2756
Vanz, A. L. , Renard, G. , Palma, M. S. , Chies, J. M. , Dalmora, S. L. , & Basso, L. A. , et al. (2008). Human granulocyte colony stimulating factor (hg-csf): cloning, overexpression, purification and characterization. Microbial Cell Factories, 7(1), 1-12.
Malekian, R., Jahanian-Najafabadi, A., Moazen, F., Ghavimi, R., Mohammadi, E., & Akbari, V. (2019). High-yield Production of Granulocyte-macrophage Colony-stimulating Factor in E. coli BL21 (DE3) By an Auto-induction Strategy. Iranian journal of pharmaceutical research : IJPR, 18(1), 469–478.
Koeninger, L. , Armbruster, N. S. , Brinch, K. S. , Kjaerulf, S. , & Wehkamp, J. . (2020). Human β-defensin 2 mediated immune modulation as treatment for experimental colitis. Frontiers in Immunology, 11, 93.
Mauro V. P. (2018). Codon Optimization in the Production of Recombinant Biotherapeutics: Potential Risks and Considerations. BioDrugs : clinical immunotherapeutics, biopharmaceuticals and gene therapy, 32(1), 69–81. https://doi.org/10.1007/s40259-018-0261-x
Chassaing, B. , Aitken, J. D. , Malleshappa, M. , & Vijay-Kumar, M. . (2014). Dextran sulfate sodium (dss)-induced colitis in mice. Curr Protoc Immunol, 104, Unit 15.25.
Zhou, Y. , Liu, P. , Gan, Y. , Sandoval, W. , Katakam, A. K. , & Reichelt, M. , et al. (2016). Enhancing full-length antibody production by signal peptide engineering. Microbial Cell Factories, 15(1).
Van Gerven, N., Klein, R. D., Hultgren, S. J., & Remaut, H. (2015). Bacterial amyloid formation: structural insights into curli biogensis. Trends in microbiology, 23(11), 693–706.
Zhao, F. , Song, Q. , Wang, B. , Du, R. , Han, Y. , & Zhou, Z. . (2018). Secretion of the recombination α-amylase in escherichia coli and purification by the gram-positive enhancer matrix (gem) particles. International Journal of Biological Macromolecules, 123.
Sriwidodo, S., Subroto, T., Maksum, I. P., Wathoni, N., Rostinawati, T., Ulya, H., & Putri, I. U. (2019). Optimization of Secreted Recombinant Human Epidermal Growth Factor Production Using Pectate Lyase B from Escherichia Coli BL21(DE3) by Central Composite Design and Its Production in High Cell Density Culture. Journal of pharmacy & bioallied sciences, 11(Suppl 4), S562–S566.
Mohajeri, A. , Abdolalizadeh, J. , Pilehvar-Soltanahmadi, Y. , Kiafar, F. , & Zarghami, N. . (2016). Expression and secretion of endostar protein by escherichia coli: optimization of culture conditions using the response surface methodology. Molecular Biotechnology, 58(10), 634-647.
Chen, L. U. , Zhao, H. , Zou, W. Y. , Fan, Q. L. , Yong-Biao, F. U. , & Song, L. H. . (2011). Secretion expression of recombinate human interferon α-2b by escherichia coli. Journal of Biology.
Isabel Guerrero㎝ontero, Richards, K. L. , Jawara, C. , Browning, D. F. , & Robinson, C. . (2019). Escherichia coli 'tatexpress' strains export several g/l human growth hormone to the periplasm by the tat pathway. Biotechnology and Bioengineering, 116(12).
Christie, P. J. . (2019). The rich tapestry of bacterial protein translocation systems. The Protein Journal, 38(M111).
Liu, J. , Kim, S. Y. , Sun, S. , Jung, S. H. , & Chung, Y. J. . (2018). Overexpression of tff3 is involved in prostate carcinogenesis via blocking mitochondria-mediated apoptosis. Experimental & Molecular Medicine, 50(8).
Riglar, D. T. , Giessen, T. W. , Baym, M. , Kerns, S. J. , Niederhuber, M. J. , & Bronson, R. T. , et al. (2017). Engineered bacteria can function in the mammalian gut long-term as live diagnostics of inflammation. Nature Biotechnology.
Da effler, K. N, Galley, J. . , Sheth, R. U. , Ortiz-Velez, L. C. , Bibb, C. O. , & Shroyer, N. F. , et al. (2017). Engineering bacterial thiosulfate and tetrathionate sensors for detecting gut inflammation. Molecular Systems Biology,13,4(2017-04-03), 13(4), 923.
Alexis, Courbet, Drew, Endy, Eric, & Renard, et al. (2015). Detection of pathological biomarkers in human clinical samples via amplifying genetic switches and logic gates. Science Translational Medicine.
Rottinghaus, A. G. , Amrofell, M. B. , & Moon, T. S. . (2020). Biosensing in smart engineered probiotics. Biotechnology Journal, 15(10).
Vlková, M., Morampalli, B. R., & Silander, O. K. (2021). Efficiency of the synthetic self-splicing RiboJ ribozyme is robust to cis- and trans-changes in genetic background. MicrobiologyOpen, 10(4), e1232.
Clifton, K. P., Jones, E. M., Paudel, S., Marken, J. P., Monette, C. E., Halleran, A. D., Epp, L., & Saha, M. S. (2018). The genetic insulator RiboJ increases expression of insulated genes. Journal of biological engineering, 12, 23.