Dr. Tagawa, one of our PIs who conducts research using in vitro models of IBD, provided the following crucial advice:

For the medical application of genetically modified probiotics, biosafety and ethical considerations are essential. Therefore, a safety mechanism such as a Kill Switch must be incorporated to prevent their spread into the external environment.

Design Direction

Following these discussions, we established three main pillars for our project:

1. Colonization Strategy Utilizing Prebiotics

   Introduce a xylitol metabolism pathway into E. coli Nissle 1917 to enable it to selectively colonize and grow by using a carbon source that other gut bacteria cannot utilize.

2. Production of Therapeutic Substances

   Treat IBD by producing substances that suppress inflammation and promote healing.

3. Introduction of an Inflammation-Dependent Kill Switch

   Design a tetrathionate-responsive Kill Switch that induces self-destruction when gut inflammation subsides, achieving both safety and self-regulation.

Why did we establish contact?

Professor Wachi has been researching molecular biology at Institute of Science Tokyo for over 30 years, studying bacterial growth, metabolism, and antibiotics. He possesses deep knowledge of the metabolic and growth mechanisms of microorganisms, including E. coli.

To create a probiotic that assimilates xylitol, we devised a plan based on literature to convert xylitol into xylulose, a sugar that E. coli can use, and merge it into the pentose phosphate pathway. For this, we intended to introduce xylitol dehydrogenase. Furthermore, we planned to introduce a xylitol transporter from Pantoea ananatis into EcN to increase xylitol uptake. We sought his opinion on the validity of this strategy.

Additionally, we were considering having the probiotic secrete the immunosuppressive cytokine IL-10. We asked for his advice on whether to choose E. coli or Gram-positive bacteria like Bacillus subtilis or lactic acid bacteria for this purpose. We also asked for suitable experimental strains.

Regarding the therapeutic substance, we were considering the secretion of the immunosuppressive cytokine IL-10 but were undecided on whether to choose E. coli or Gram-positive bacteria (B. subtilis, lactic acid bacteria) as the host. We also sought advice on selecting appropriate experimental strains to proceed with these experiments. notion

What did we learn?

From our conversation with Professor Wachi, we gained important insights into each technical challenge. He introduced existing research showing that xylose transporters can uptake xylitol to some extent. In that study, by overexpressing E. coli’s native propanediol dehydrogenase, they successfully dehydrogenated xylitol and merged it with downstream metabolic pathways [1]. However, the doubling time was 10 hours, and the efficiency was not at a practical level. This case suggested that with proper engineering of the downstream metabolic pathways, xylitol utilization is possible in principle.

He also mentioned that introducing a transporter could improve xylitol uptake. He speculated that since Pantoea ananatis is a close relative of E. coli, the compatibility for expression is likely high. Nevertheless, he pointed out that concerns about efficiency and size compatibility would remain if introduced, and the actual behavior would need to be verified experimentally.

Furthermore, regarding IL-10 secretion, it was reconfirmed that the outer membrane of E. coli poses a major barrier. Methods like co-expression with outer membrane proteins or using membrane vesicles are theoretically possible but not well-established. In contrast, Gram-positive bacteria have simpler secretion mechanisms and can be expected to secrete efficiently. However, B. subtilis and Bacillus natto have difficulty colonizing the gut, presenting a significant trade-off. We face a choice: prioritize gut colonization with E. coli, or prioritize high-level secretion with Gram-positive bacteria.

Finally, regarding the Kill Switch, Professor Wachi pointed out that making it solely dependent on xylitol by restricting the probiotic’s carbon source to xylitol is difficult. Microorganisms utilize many types of carbon sources, and we have yet to identify all of them. Thus, identifying and blocking all of them is impossible with current technology. As an alternative, he suggested that a more realistic approach for the Kill Switch would be to redesign E. coli’s native systems, such as the toxin-antitoxin system.

We also learned that the gut environment is not completely anaerobic; it can be partially aerobic, especially in inflamed areas. This needs to be considered when selecting and designing strains.

How did it impact our project?

This discussion led to several directional changes in our project.

• We will take a phased approach: use easily transformable strains at the experimental level and then move to probiotic strains for the final application.
• Regarding IL-10 secretion, since secretion from EcN is difficult, we decided to proceed with parallel studies in B. subtilis and lactic acid bacteria.
• Our Kill Switch design policy was also refined. We realized that blocking all carbon utilization pathways other than xylitol was impractical. We abandoned this initial idea and decided to design the Kill Switch by repurposing E. coli’s native toxin-antitoxin system.

Through this interview, we were able to establish concrete design pillars—“colonization aid via xylitol utilization,” “host selection for IL-10 secretion,” and “toxin-antitoxin-based Kill Switch design”—which allowed us to revise our project plan to be more realistic and advanced.

Why did we establish contact?

Dr. Hirasawa is an associate professor at Institute of Science Tokyo, where he researches the production of useful substances by microorganisms through metabolic engineering and microbial breeding. His approach is rich in Dry Lab knowledge, including not only synthetic biology techniques but also the use of omics data analysis and metabolic simulations.

When designing the xylitol assimilation pathway, we were concerned that simply expressing xylitol dehydrogenase might not achieve a sufficient growth rate, as it would channel xylulose, a sugar not typically used in large amounts, into the pentose phosphate pathway. One reason for this concern was the very long doubling time of the E. coli that was engineered to utilize xylitol in the interview with Professor Wachi. To address this, we planned to decide which enzymes, other than xylitol dehydrogenase, should be produced based on Dry Lab simulations and Human Practices. Therefore, we were introduced to Dr. Hirasawa by Professor Wachi and interviewed him.

What did we learn?

Through our discussion with Dr. Hirasawa, we gained several key insights. First, he pointed out that the initial step of xylitol uptake depends on the transporter’s characteristics, and expressing xylitol dehydrogenase alone may not provide sufficient metabolic flux. In particular, the key is whether E. coli’s native xylose ABC transporter or a newly introduced xylitol transporter can transport xylitol. He stated that this cannot be determined by models alone and requires experimental verification. It is possible to introduce about 4 kb into a plasmid, so expressing and testing it is the most reliable method. He advised that if we want to investigate sugar binding to the transporter, we should consult a protein engineering specialist.

Dr. Hirasawa noted that xylitol metabolism is not just about producing xylulose-5-phosphate; the activity of multiple enzymes in the pentose phosphate pathway could be rate-limiting. In fact, Dry Lab simulations show that multiple reactions, including TKT1 (transketolase), are heavily utilized, and optimizing the metabolic balance requires the co-expression and adjustment of multiple enzymes. He also mentioned that factors like oxygen and the NADH/NADPH balance could have a significant impact. He advised that a simple Flux Balance Analysis (FBA) scheme is insufficient, and more detailed enzyme-level simulations would be effective if necessary.

How did it impact our project?

• While E. coli might be able to uptake xylitol on its own, we decided to newly introduce a xylitol transporter and test it experimentally.
• Recognizing that simply introducing xylitol dehydrogenase might be insufficient, we decided to introduce enzymes like TKT1 as candidates, based on Dry Lab experiments and Human Practices, considering the balance of NADH, NADPH, and protons. The specific enzyme selection will be made considering various factors, such as the load on the plasmid.
• We reaffirmed the need to combine verification from both Dry and Wet Lab perspectives and gained an understanding from a Human Practices standpoint that “metabolic pathway design is not just about introducing a single gene but about optimizing the overall flux balance.”

Why did we establish contact?

Dr. Ohno is a lecturer at Institute of Science Tokyo researching systems biology of metabolism and is involved in diagnosing and treating diseases, including IBD, from a systems biology perspective.

Based on our interview with Dr. Hirasawa, we decided to introduce enzymes like TKT1 as candidates, considering the balance of NADH, NADPH, and protons from Dry Lab and Human Practices perspectives to increase the success rate of the xylitol utilization pathway. We sought Dr. Ohno’s opinion, who has deep knowledge of IBD and analysis, to finalize the specific model design.

What did we learn?

Through our discussion with Dr. Ohno, we understood several essential challenges in designing the xylitol metabolic pathway. It was pointed out that conventional Flux Balance Analysis (FBA) is too simplified to accurately reflect the rate-limiting enzymes and expression level constraints within a real cell. Therefore, we learned that to perform more realistic simulations, it is necessary to use methods like an enzyme-constrained model or MPEK, which incorporate parameters such as enzyme Kcat and expression levels.

We also gained insights into the redox balance when converting xylitol to xylose. Dr. Ohno advised that if NADP+ or NAD+ becomes deficient in the cell, xylitol is likely to accumulate, risking a stall in the overall metabolism. Therefore, it would be better to introduce enzymes that regenerate NADP+ or NAD+. He also introduced prior research reporting that NAD-dependent pathways are more favorable for biomass production. This led us to understand that when selecting a xylitol dehydrogenase, we should choose an NAD-utilizing enzyme.

Additionally, he stated that catabolite repression is an element that cannot be ignored when considering the gut’s metabolic environment. E. coli will always prioritize glucose if it is present. Therefore, our design’s advantage was confirmed, as a strain capable of using xylitol would have a nutritional niche in the gut environment.

How did it impact our project?

• Understanding that simple FBA-based simulations cannot correctly evaluate metabolic pathway bottlenecks, we revised our policy to use enzyme-constrained models for more realistic metabolic design. This will enable us to identify rate-limiting enzymes and formulate clear experimental strategies to improve them.
• We decided to incorporate a new strategy to convert the NADPH dependency of xylitol dehydrogenase to NAD dependency. This has the potential to improve the coenzyme balance and make biomass production more efficient.
• Considering the phenomenon of glucose priority utilization in the gut, it was reaffirmed that our designed xylitol utilization system has the strength of securing a metabolic niche not utilized by existing carbon sources.
• Overall, the discussion with Dr. Ohno was a catalyst that advanced our research from simple pathway introduction to a more sophisticated metabolic design incorporating coenzyme dependency optimization and enzyme modification.

Why did we establish contact?

Professor Taguchi researches the 3D structure of proteins at Institute of Science Tokyo. He utilizes model organisms like E. coli and yeast, and performs functional analysis and imaging of purified proteins, possessing deep knowledge of protein production and analysis using microorganisms.

For the production of IL-10, we had to abandon the use of B. subtilis and lactic acid bacteria due to the limitations of the strains we could handle, and decided to produce IL-10 in E. coli. However, IL-10 has disulfide bonds, making proper folding difficult, and secretion volume after transport to the periplasm is low. Therefore, we sought his opinion on suitable strains for IL-10 production, measurement methods, and secretion methods.

What did we learn?

Through discussions with Professor Taguchi, we reaffirmed that there are several major technical hurdles in producing and secreting IL-10 in E. coli. The selection of signal peptides (e.g., DsbA, TorT, TolB) using the Sec or SRP-dependent pathways is crucial, and it’s necessary to experimentally compare multiple candidates.

Furthermore, Professor Taguchi proposed strains like Origami and Rosetta-gami as a way to enhance secretion. These strains have an oxidative intracellular environment, which facilitates the formation of disulfide bonds, making them potentially well-suited for IL-10 production. He also suggested a strategy of using “leaky strains” with increased outer membrane permeability to improve secretion from the periplasm. The L-form strain, which lacks a cell wall and has reduced outer membrane protein expression, increases secretion. Leaky strains are those in which cell wall synthesis enzymes or outer membrane synthesis enzymes have been removed. Using these could increase the secretion of IL-10 that has reached the periplasm. However, Professor Taguchi warned that leaky strains are not selective in what they secrete and are fragile, making them potentially unsuitable for long-term use in the gut, though they might be viable for industrial applications.

He also advised that to detect secretion into the medium, besides antibodies, systems like the split-GFP system could be used. In this system, part of GFP is released into the medium, and another part is attached to the tail of the secreted IL-10. When IL-10 is released into the medium, its presence can be quantified by fluorescence intensity. This measurement method could be an option if the secretion pathway is compatible.

For our project’s future design, Professor Taguchi proposed a strategy of genetically editing native gut bacteria using CRISPR. This would increase the chances of solving the colonization challenge and lead to effective treatment.

How did it impact our project?

• We determined that using B. subtilis or lactic acid bacteria was difficult with our technical resources and confirmed our policy to use E. coli as the base. Among them, we decided to prioritize testing conditions for correct IL-10 folding using Origami-series or Rosetta-gami strains, which enable disulfide bond formation.
• To confirm successful secretion, we will consider the innovative method of split-GFP screening in addition to conventional antibody-based detection.
• Since uncertainties remain regarding secretion efficiency and cell stability in the gut environment, we decided to investigate multiple approaches in parallel.

Why did we establish contact?

Professor Hayashi is one of our team’s PIs and researches protein function and antibodies. He applies High-Performance Proteomics methods, which can comprehensively and universally study proteins in cells, to understand microbial and cellular phenomena, and applies this knowledge to disease diagnosis, treatment, and biotechnology development.

We sought advice from Professor Hayashi, who has deep knowledge of protein analysis, on the validity of our detailed experimental conditions and plans for producing and secreting IL-10 in E. coli.

What did we learn?

Professor Hayashi emphasized that the disulfide bond (S–S bond) in IL-10 is not essential for its structure formation itself but plays a role in stabilizing the dimeric structure. IL-10 folds while constantly fluctuating and is then fixed by the S–S bond. He pointed out that a major cause of misfolding is that when multiple molecules are synthesized at once, they can incorrectly form S–S bonds with nearby molecules before proper folding is complete. He advised that high expression with a strong promoter or high-temperature cultivation is prone to causing aggregation, which could be improved by using a weak promoter or low-temperature cultivation.

Furthermore, we learned that “whether it exerts physiological activity” is more important than structural integrity. If it remains soluble and does not aggregate, the next step should be to confirm its activity using an assay with immune cells. He also advised that tags for purification and detection could interfere with activity, so it’s necessary to consider the design, such as having tagged and untagged versions for concentration detection and effect verification.

How did it impact our project?

• We revised our perspective to prioritize “confirming function” over “confirming structural integrity.” As a result, rather than relying on special strains (e.g., Origami strains), we decided to first adopt weaker expression conditions (mild promoter, low-temperature cultivation) in standard E. coli strains to minimize aggregation and misfolding.
• Furthermore, we incorporated a biological activity assessment of IL-10 using immune cells into our plan and decided on a policy to create two types of strains for tag design: one with a tag for concentration measurement and one without a tag for activity assessment.
• Ultimately, due to the high risk of misfolding from its large molecular weight and dimeric structure, we decided not to proceed with IL-10 production. However, the advice we received here has been applied to the secretion of EGF, such as creating tagged and untagged strains.

Why did we establish contact?

Professor Tsuchiya is a clinical physician at the forefront of IBD treatment as the director of the University of Tsukuba IBD Center. He actively incorporates the latest treatments with high expertise while also being involved in training young doctors, working with a view toward the future of IBD treatment. A key feature is his academic background as a researcher, allowing him to study the pathology of IBD from a unique perspective, utilizing both basic research and clinical data.

Through previous interviews and literature review, we had determined that producing IL-10 in EcN was technically difficult. We selected alternative candidate substances from the literature and sought Professor Tsuchiya’s opinion, who has deep knowledge of current and future required treatments, on what would be the most appropriate substance for EcN to produce. Additionally, we asked for his perspective on how much our project could contribute to the current treatment system.

What did we learn?

In our interview with Professor Tsuchiya, we deepened our understanding of several concepts crucial for advancing our project.

First, he pointed out that many current clinical treatments rely on immunosuppressants. These drugs work by suppressing the pathway where immune cells are activated, produce inflammatory cytokines, and cause inflammation in the intestinal tissue. However, while they can suppress symptoms, they do not lead to a fundamental cure, and the remission rate remains at about 50%.

Furthermore, Professor Tsuchiya explained the important concept of the “Vicious cycle of Inflammation.” The gut’s mucosal surface has a “mucus layer” like a moat, and inside that, there is a “mucosal epithelial cell” layer like a castle wall. If this moat is well-maintained, bacteria and foreign substances cannot easily enter. However, if the mucus layer thins due to inflammation or damage, and the epithelial cell wall collapses, bacteria-derived components and endogenous damage-associated molecular patterns (DAMPs) invade the body. Then, the immune cells, acting as soldiers, overreact to the “enemy attack,” intensifying the inflammatory response. As a result, the castle wall is further destroyed, the moat cannot be maintained, and the cycle of invasion and inflammation repeats. Thus, after the mucus layer is stripped, and epithelial cell damage progresses, various foreign substances invade the mucosa, further activating the immune system and creating a “Vicious cycle of Inflammation.”

Against this backdrop, it has become very difficult to achieve remission in IBD with a single drug. However, if multiple current immunosuppressants are used concurrently, they suppress the immune system too strongly, leading to significant side effects, so their combined use is limited. On the other hand, if a drug has a different mechanism of action from immunosuppression, treatment from multiple mechanisms could enhance therapeutic effects while suppressing side effects. From this point, Professor Tsuchiya stated that there is a high need for treatments that can be used in combination with other drugs.

Professor Tsuchiya provided the following evaluation for the three candidate substances we selected from literature review for EcN to produce:

From this feedback, we determined that EGF best meets the clinical needs.

Professor Tsuchiya also shared valuable insights regarding the development of therapies focused on epithelial cell repair and regeneration. In the past, research was conducted to induce epithelial cell proliferation and wound healing using Hepatocyte Growth Factor (HGF). Although HGF has been reported to promote intestinal epithelial recovery, its effects were generally achieved through systemic administration, such as intravenous drips. However, with systemic administration, the amount of HGF that passes from the blood through the intestinal wall to reach the epithelial cells is limited, posing a major constraint on its practical application as a drug. In contrast, Professor Tsuchiya pointed out that having EcN produce EGF directly in the gut is highly likely to be effective. Also, therapies aimed at healing epithelial cells have high hurdles for commercialization due to costs, so pharmaceutical companies are currently focused on developing immunosuppressants. Therefore, if our project is commercialized, the cost could be kept low as it is bacteria-derived, and it could become an excellent drug that can be used in combination with standard treatments.

Through our literature review and interviews, we were concerned that if the secretion amount of the substance produced by EcN was insufficient, it might not have a significant effect. However, Professor Tsuchiya mentioned that when EcN dies and lyses in the gut, the therapeutic substance accumulated inside the cells could be released all at once. Therefore, it might be effective even with a low secretion rate.

Furthermore, Professor Tsuchiya also emphasized the importance of medication adherence. He introduced a paper he often uses in his lectures [4] and gave an example of a drug improvement made from the perspective of medication adherence. One of the basic drugs, mesalazine (Asacol), required taking 3 capsules 3 times a day, and forgotten doses were a problem. To solve this, Lialda was released, which is taken as 4 tablets once a day, reducing missed doses. From this, we believed that our colonization support system, which focuses on reducing the frequency of medication, is worth pursuing.

Professor Tsuchiya pointed out that although probiotics are commonly prescribed, they rarely colonize. It is generally very difficult for foreign microorganisms to colonize the gut. Therefore, he suggested that a system that assists colonization by consuming xylitol along with EcN, allowing it to assimilate xylitol, would be highly effective in cases like this where the benefit of EcN colonizing the gut is significant. He also believes that the negative effects of xylitol intake on the gut are not that significant, and the benefit of stopping inflammation outweighs the harm of taking xylitol.

On the other hand, Professor Tsuchiya pointed out that the colonization of EcN in the gut could become a problem if a patient wants to stop taking the drug but cannot. Therefore, he agreed that the we are developing is a very effective measure. As a practical issue, he also made the important point that when administering our developed “xyego” to Japanese people, who often use antibiotics for treatment, the colonized bacteria might frequently die, so the intake schedule must be carefully considered.

How did it impact our project?

• We confirmed that achieving remission in IBD is difficult due to the “Vicious cycle of inflammation” and that a combination of multiple therapies is necessary.
• We adopted the policy of having EcN produce EGF, as epithelial cell repair with EGF meets this unmet need.
• The point that “probiotics have difficulty colonizing the gut long-term” supported the validity of our design to introduce a xylitol assimilation pathway.
• We reconfirmed that the design of a Kill Switch, which would be suicidally removed after treatment, ensures safety.
• We were able to realize the process of Integrated Human Practices by integrating feedback from a clinical perspective with basic research data and reflecting it in our project design.

Why did we establish contact?

We interviewed Professor Wachi again. We had decided in a previous interview to use a toxin-antitoxin system for the Kill Switch design, but we couldn’t determine the details of the timing and strength of toxin and antitoxin expression, so we sought advice to create the design. Furthermore, in a previous discussion with our PIs, we decided to create a system to remove the antibiotic resistance gene to prevent its spread in the gut, and we asked about the feasibility of this system.

What did we learn?

Through our discussion with Professor Wachi, we gained several important insights regarding the Kill Switch design. First, he explained that the MazE/MazF system is conserved on the chromosome. The mechanism is that MazE and MazF are expressed simultaneously, and when expression stops due to stress like nutrient starvation, MazE, being more unstable than MazF, degrades first, leaving only MazF, which leads to cell death. Many toxin-antitoxin systems work similarly and are used for plasmid maintenance. He suggested that when we express them, it would be good to express the toxin and antitoxin simultaneously.

It was also pointed out that while tetrathionate is expected to be sufficiently produced during inflammation in the gut, the actual concentration in the human gut depends heavily on the severity of inflammation and has not been determined. This was recognized as an uncertainty in evaluating the clinical effectiveness of our system.

Furthermore, regarding the external addition of tetrathionate, he believed it could be mixed into research media without issue, but its potential effects on E. coli are unknown. He advised that it might be necessary to design the system so that MazF can be controlled by an external inducer (e.g., IPTG) in some cases.

For the removal of the antibiotic resistance gene, he suggested that simple Cre-loxP excision risks releasing DNA fragments into the environment. As an alternative, he proposed the introduction of an auxotrophic marker (e.g., disruption of a thymine or vitamin synthesis gene). This would be a very reliable Kill Switch in terms of gene disruption but would require consideration of the difficulty of gene introduction and the potential decrease in colonization efficiency.

Furthermore, Professor Wachi proposed the Restriction-Modification system (R-M system) as a system similar to the toxin-antitoxin system that could remove antibiotic resistance. In this system, a restriction enzyme and a modification enzyme are expressed simultaneously. The modification enzyme modifies the DNA so that the restriction enzyme cannot cut it. If the expression of this system stops for any reason, such as plasmid loss, the modification enzyme degrades, DNA can no longer be modified during replication, and the restriction enzyme cuts the DNA, leading to cell death. Depending on the choice of restriction enzyme, this system can cut the antibiotic resistance gene, allowing for the removal of antibiotic resistance simultaneously with the Kill Switch function. Therefore, we decided to primarily use this system in our future designs.

How did it impact our project?

• We decided to focus the design of our Kill Switch on a tetrathionate-inducible R-M system.
• We learned that the R-M system is a design that fits our concept, as it can simultaneously perform the functions of a Kill Switch and antibiotic resistance gene removal.

Why did we establish contact?

Dr. Fukahori serves as a specially appointed lecturer at Institute of Science Tokyo. Coming from a pharmaceutical company background, he is involved in activities to create innovation through industry-academia collaboration. We sought his opinions on the concrete details of our project’s final design, potential risks in its use, and other aspects of social implementation.

What did we learn?

Dr. Fukahori pointed out that modern medicine is moving towards personalized medicine. As many stakeholders have noted, there is a demand for treatments targeting people with specific attributes rather than a one-size-fits-all approach.

He pointed out that the diversity of gut bacteria is strongly related to the patient’s treatment efficacy and colonization. While our initial plan to introduce a gene into a single type of E. coli is simple and experimentally manageable, the gut environment differs for each patient. Therefore, introducing a similar system into multiple bacterial species could make it more likely to colonize by matching each patient’s specific environment. He shared an example from a real business where transplanting 15 types of gut bacteria was shown to activate the gut. This aligns with findings from fecal microbiota transplantation research, where maintaining diversity is considered important.

Dr. Fukahori also noted that complex conditions like IBD are difficult to cure with a single intervention, and combination with other therapies is a prerequisite. In that respect, the fact that this project’s approach is easily combinable with other treatments is a major advantage.

On the other hand, he warned that if the bacteria proliferate excessively, the amount of EGF could become excessive and pose a risk. Excessive EGF can also cause cell damage. Therefore, we felt it necessary to investigate the amount of EGF that could become a risk.

How did it impact our project?

• This led us to consider expanding the project’s final design from administering only EcN to a system with multiple strains. This provides a direction for a flexible design that can be adapted to each patient’s gut environment while maintaining diversity.
• The point about the risk of excessive EGF administration motivated us to place even greater emphasis on safety design. Therefore, we now think it is necessary to investigate the concentration at which EGF has adverse effects and to consider countermeasures if such effects occur.
• These insights provided an important foundation for designing a probiotic therapy that combines reliability and safety for future social implementation, moving beyond just basic research.

Why did we establish contact?

Dr. Yamada is an associate professor at Institute of Science Tokyo, where he researches gut microbiota analysis using bioinformatics. He has also launched a startup company utilizing gut microbiota, engaging in activities that apply this knowledge to society, from fecal analysis and healthcare to medical applications. We sought his opinions on potential implementation strategies and the hurdles we might face when bringing our project to society.

What did we learn?

Through our discussion with Dr. Yamada, we gained important insights into the feasibility of social implementation and the biological conditions for success.

First, Dr. Yamada pointed out that “it is difficult to make foreign gut bacteria colonize using only xylitol.” Long-term colonization in the gut requires multiple conditions, not just nutrient utilization, but also adhesion factors, immune evasion, and interactions with other bacteria. Therefore, we realized that simply introducing a xylitol metabolism system might not satisfy all the factors for colonization. This made us feel the need to confirm whether the strain we use can meet these other factors during selection.

Regarding EGF secretion, he emphasized the need to confirm in vitro whether the protein is actually being released outside the cell. If our designed E. coli can be shown to secrete EGF at a higher concentration than previously reported lactic acid bacteria, it could be a significant advantage for therapeutic application.

Furthermore, Dr. Yamada also touched upon the issue of the Cartagena Protocol in social implementation. He mentioned that using phages to genetically modify native gut bacteria might be a way to circumvent the Cartagena Protocol. While the technical hurdles to achieve this are high, if realized, it could allow for treatment with high colonization efficiency.

Dr. Yamada also pointed out that if the goal is the treatment itself, there are ways to achieve it without genetic modification. From this perspective, he suggested that we need to reconsider the justification for our project requiring genetic modification.

Dr. Yamada also noted that xylitol carries a risk of osmotic diarrhea. Therefore, we felt it necessary to reconfirm that the dosage of xylitol would not cause adverse effects.

How did it impact our project?

• Recognizing that xylitol metabolism alone is insufficient for adequate colonization, we felt the need to confirm that our selected strain meets other colonization factors.
• We planned an experiment to confirm in vitro that EGF is actually secreted outside the cell. Specifically, we aim to secrete a therapeutically effective concentration, referencing existing lactic acid bacteria studies.
• Our design may fall under the Cartagena Protocol’s definition of an “open-system genetically modified organism.” To avoid this, we began to consider a final design that utilizes bacteriophages for transformation.

Dr. Taku Kobayashi

tags: Colonization Aid, Mucosal Healing, Future Design

Why did we establish contact?

Dr. Kobayashi is an associate professor at Kitasato University and the director of the Inflammatory Bowel Disease Advanced Treatment Center at the university hospital. Dr. Kobayashi is involved with patients on the front lines by continuing to provide the latest IBD treatments and advocates for a treatment policy that emphasizes patient Quality of Life (QOL). He selects methods for tests and surgeries that are as minimally invasive as possible and chooses drugs tailored to individual circumstances, focusing not only on efficacy but also on convenience. We conducted an interview with Dr. Kobayashi to seek his opinions on the advantages and concerns of implementing this project in society at the stage of deciding our project’s design policy.

What did we learn?

Through our interview with Dr. Kobayashi, we were able to gain important insights into the current state of IBD treatment.

Currently, in addition to steroids and immunosuppressants, various antibody drugs and small molecule drugs such as TNF inhibitors, IL-12/23 inhibitors, JAK inhibitors, and integrin inhibitors are already used in clinical practice, each controlling the immune system through different mechanisms of action. Dr. Kobayashi emphasized that efforts are underway to further increase the therapeutic effects of current treatments.

Dr. Kobayashi agreed that to stop the “Vicious cycle of inflammation,” it is better not only to stop the inflammation with immunosuppressants but also to heal the mucosa and stop the cycle from both directions. Therefore, the high demand for our approach of producing EGF was confirmed. He also stated that this therapy is not a systemic treatment like an injection but one that acts only in the gut, so the risk of side effects is low, and it can be used long-term.

Dr. Kobayashi also pointed out that personalized medicine has been gaining importance in recent years. There are now many treatment options, and the effective drug differs for each individual patient. Therefore, he highly praised our project for being able to be combined with other treatments and for increasing the range of treatment options. He also mentioned that by advancing our “Xylego,” it might become possible to perform a treatment that originally required multiple drugs with just “Xylego” alone, by having it produce both immunosuppressive and epithelial-healing proteins.

On the other hand, Dr. Kobayashi also presented several important concerns. First is the relationship with anticancer drugs that are EGF inhibitors. Some anticancer drugs inhibit EGFR to suppress cancer cell growth, and in this case, administering EGF might have a very small effect.

Furthermore, regarding xylitol use, he pointed out that since it is a type of Fermentable Oligosaccharide, Disaccharide, Monosaccharide, and Polyol (FODMAP), which is problematic for patients with Irritable Bowel Syndrome, excessive intake carries a risk of diarrhea and abdominal bloating. We felt it necessary to carefully consider the conditions so that xylitol intake does not become excessive. In a subsequent literature review, we obtained data suggesting that an intake of less than 15g-20g at one time is unlikely to cause adverse effects.

Dr. Kobayashi pointed out that while it has been confirmed that the strain used for verification, BL21(DE3), lacks the ability to metabolize xylitol, it is necessary to reconfirm whether E. coli Nissle’s metabolic pathway truly lacks xylitol metabolism capability.

Finally, we also discussed the social hurdles of administering genetically modified microorganisms into the body. He stated that if this hurdle is to be overcome, a design that uses a cocktail of multiple bacterial strains would be more clinically persuasive than a single strain.

How did it impact our project?

• We confirmed that the use of EGF is clinically valid in terms of “epithelial repair” and is highly likely to be suitable for long-term use.
• Regarding xylitol, we began to consider the risk as a FODMAP at the design stage, realizing the need for adjustments according to each patient’s symptoms and tolerance. This shifted our metabolic system design to consider not only “improving gut colonization” but also “ensuring patient safety.”
• We decided to consider a “cocktail of multiple strains” as our final treatment strategy.
• The dialogue with Dr. Kobayashi was a major catalyst in evolving our project from a lab-based study to a “realistic solution that can reach the clinic.”

Why did we establish contact?

We planned to investigate the number of xylitol-utilizing bacteria and their impact using a Dry Lab approach and sought advice from Dr. Yamada, who has excellent knowledge of gut microbiota analysis.

What did we learn?

Dr. Yamada first questioned the premise of our project—“treating IBD by colonizing the gut with EGF-producing bacteria”—asking if the scientific evidence was sufficient. He stressed that we should investigate papers that actually show a correlation between “bacterial colonization rate” and “therapeutic effect” to strengthen the validity of our hypothesis.

He also emphasized the importance of clearly understanding the difference between the “engraftment-type” strategy we were aiming for and the “transient-type” strategy adopted by many probiotics. The former colonizes almost permanently and exerts its effect. The latter deliberately avoids long-term colonization and improves the gut environment through a short-term stay, which is more socially accepted due to its safety and controllability. From this comparison, we felt the need to verify whether a short-term stay could also be effective, without insisting on “complete colonization.”

Based on Dr. Yamada’s suggestion, we planned to work on quantifying the colonization modeling. The plan is as follows:

1. Database Analysis:
   • Use KEGG, HMP, curatedMetagenomicData, etc., to comprehensively analyze the types and proportions of gut bacteria possessing the xylitol metabolism gene (e.g., K00006).
   • This will quantify “how many xylitol-utilizing bacteria exist in the gut.”
2. Colonization Model Construction:
   • Model ①: Condition where no xylitol-utilizing bacteria are present in the gut.
   • Model ②: Condition where xylitol-utilizing bacteria are present.
   • Compare the growth, competition, and colonization of foreign bacteria under these two conditions. Calculate using the formula: (xylitol utilization per bacterium) × (abundance ratio of utilizing bacteria) as a parameter. Finally, model the competition with resident bacteria that use glucose (survival competition after colonization).
3. Model Accuracy Verification:
   • Evaluate the model’s validity by comparing the simulation results with the results of co-culture experiments conducted in the Wet Lab.

How did it impact our project?

• We reconfirmed the importance of searching for papers and other sources to support the effects of EGF.
• Based on the difference between “engraftment-type” and “transient-type,” we will consider whether a transient type can also be effective.
• We decided to proceed with the colonization modeling in the order of database analysis, model construction, and comparison with Wet Lab results.

Why did we establish contact?

Dr. Hashimoto is an associate professor of molecular biology at National Cheng Kung University in Taiwan. He is also one of the PIs for the NCKU iGEM team and is knowledgeable about experimental planning, common failures, and their solutions from the perspective of an iGEM team. We went to Dr. Hashimoto for advice on our two-plasmid transformation, which was repeatedly failing, and for an evaluation of our future experimental plans.

What did we learn?

Our transformation protocol involved first transforming BL21(DE3) with one plasmid, then making that strain into new competent cells. We then transformed the competent cells containing one plasmid with another plasmid to create a strain with two plasmids. However, the transformation of the second plasmid after making competent cells was failing. When we asked for advice on this, Dr. Hashimoto suggested that it is better to transform both plasmids at the same time. He said that transformation with an assembled plasmid requires high transformation efficiency because the plasmid amount is very small, but when using already purified and concentrated plasmids, high transformation efficiency is not necessary, so it is suitable to transform a single strain with an increased amount of plasmids.

He also gave advice on the Kill Switch verification system. Dr. Hashimoto pointed out that the Kill Switch is activated by DNA cleavage by a restriction enzyme. Therefore, it is unlikely that the dead cells will lyse. He stated that to confirm whether the Kill Switch has been activated, it would be best to measure the CFU (Colony Forming Units).

How did it impact our project?

• We adopted the method of transforming both plasmids into BL21(DE3) simultaneously. This resulted in a significant reduction in cloning time, allowing more time for verification experiments.
• We planned to use CFU measurement for verifying the activation of the Kill Switch. This allowed us to measure the number of viable cells more accurately than by measuring OD600.

Why did we establish contact?

We were trying to identify bottlenecks in xylitol metabolism using the ecFBA method in our Dry Lab. However, as the project progressed, several questions arose, so we went to Dr. Ohno again for advice on detailed implementation plans.

What did we learn?

Through discussions with the expert, we learned the necessity of fundamentally reviewing our modeling strategy to obtain reliable predictions adapted to the target gut environment, which is a low-oxygen, anaerobic condition.

Dr. Ohno pointed out that in a situation where ideal data for a xylitol carbon source under low-oxygen conditions is unavailable, the existing “glucose carbon source, aerobic condition” data would have a large discrepancy with the actual state of xylitol metabolism. The proteome data for a “xylose carbon source, aerobic condition,” which is considered closer to the xylitol catabolic pathway, is the most reasonable alternative at present.

Dr. Ohno stated that it is experimentally difficult to strictly determine the “oxygen uptake flux” value required to reproduce the low-oxygen conditions reflecting the gut environment in a model. He advised that if we could measure the concentration of lactate in the culture medium in the Wet Lab, we could determine if fermentation is occurring, and we decided to incorporate this into our experimental plan, considering the time available.

In the Wet Lab, xylitol assimilation was confirmed even without introducing the planned ABC transporter, which led to the hypothesis that the native xylose transporter in E. coli is likely responsible for the uptake. However, since it is thought that this alone cannot uptake a sufficient amount of xylitol, the transport mechanism is unknown. Dr. Ohno advised that instead of relying on a single hypothesis, we should compare possible mechanisms such as simple diffusion, proton symport, and ABC transport, and use this to consider the final bottleneck.

We recognized the limitation that complex physiological phenomena involving gene expression regulation, such as the expression switch from cytochrome oxidase BO3 to BD under microaerobic conditions, cannot be accurately reproduced by the ecFBA model. We learned that we need to keep this limitation in mind when interpreting the analysis results.

How did it impact our project?

• It became clear that our previous rate-limiting analysis results based on aerobic conditions likely do not reflect the target gut environment. Therefore, we decided to re-run the rate-limiting step identification simulation by combining xylose carbon source data with low-oxygen condition settings.
• A policy was established to proceed with the simulation by gradually lowering the upper limit of the oxygen uptake flux.
• To evaluate the impact of the xylitol transport mechanism, we decided to incorporate scenarios of simple diffusion and proton symport into the model, in addition to the initial ABC transporter model, and to conduct a comparison to quantitatively assess how the conclusion on the rate-limiting step depends on each transport mechanism.
• We replaced the proteome data used for the analysis from the glucose carbon source to the metabolically more similar xylose carbon source.
• We decided to interpret the results considering the inherent limitations of the model when publishing our findings.