Introduction

To realize this project, we addressed the challenges through a more engineering-driven process. Here, we employed the DBTL cycle and approached the problem from four perspectives: Design, Build, Test, and Learn.

Colonization support

Cycle1

Design

For Xylego to stably colonize the gut, it must outcompete other resident gut microbes for nutrients. To achieve this, we devised a system that partitions a metabolic niche by assimilating xylitol—a carbon source scarcely utilized by other bacteria—thereby conferring a competitive advantage.

Build

We designed a system that takes up and metabolizes xylitol and channels it into the bacterium’s pentose phosphate pathway. Specifically, we express a xylitol transporter on the cell surface and introduce xylitol dehydrogenase and D-xylulokinase so that xylitol is converted to xylulose-5-phosphate and fed into the pentose phosphate pathway. (See Design for details.)

Test

We interviewed a microbiologist and a clinician specializing in intestinal diseases about our proposed approach. (See Integrated Human Practices for details.)

Learn

From these interviews, we learned that this system is feasible. We also learned that E. coli natively possesses a xylose transporter, which may promiscuously transport xylitol.[3]

Cycle2

Design

Based on the insights obtained in Cycle1, we designed plasmids selecting xyB, xytC, xytD, and xytE present in the xyt operon derived from P.ananatis, which is closely related and can import and metabolize xylitol [4].

Build

Using Hifi assembly [5], we constructed the following plasmids and introduced them into E. coli BL21(DE3).

Test

Cloning was successful, and we obtained gel electrophoresis results.

Learn

The electrophoresis indicated that two plasmids had been introduced into E. coli. In addition, whereas we had previously introduced two plasmids by performing transformations one plasmid type at a time, when we introduced two types simultaneously, we found that two promoters could be introduced more efficiently than making the cells competent each time a plasmid was introduced.

Cycle3

design

To verify whether the E. coli constructed in Cycle2 can live using xylitol as a carbon source, we planned the following experiment.

Build

Before seeding onto M9 medium, expression was induced by adding IPTG to 0.5mM and incubating for 3 hours. Each strain was then seeded onto M9 medium containing only glucose or only xylitol as the carbon source, and OD600 was measured after 20 hours and 72 hours of incubation.

Test

The strain into which only the xylitol-metabolizing enzymes were introduced showed growth on medium with xylitol as the sole carbon source. Moreover, the strain expressing both the transporter and the metabolic enzymes showed little growth on either glucose or xylitol, whereas the strain expressing only the metabolic enzymes showed substantial growth on medium containing only xylitol.

Learn

Because the strains without transporter expression grew better, it was decided to conduct subsequent validation experiments using the strains that do not express the transporter.

It is considered that the strains expressing the transporter experienced a high metabolic burden and were unable to grow. In contrast, the strains expressing only xylitol metabolic enzymes grew sufficiently even without transporter expression, suggesting that E. coli’s native xylose transport systems or other transporters likely took up xylitol.

Growth was also observed in the strain carrying only the empty pTf16 vector even in xylitol-only medium, suggesting that the carryover of LB medium used during induction may have contributed. Therefore, it was found that washing the cells with M9 medium without any sugar prior to inoculation is necessary.

Additionally, a white precipitate was observed in samples that showed poor growth.

Cycle4

design

Since a precipitate formed in the tubes during Cycle 3, it was considered necessary to conduct an experiment to identify the composition of this precipitate.

Build

Because the precipitate was not present at the beginning of cultivation and appeared during culture, it was hypothesized that phosphate in the medium formed a complex with metal ions. Based on the medium composition, the precipitate was suspected to be a complex of magnesium and calcium ions. Therefore, an experiment was planned to dissolve the precipitate using EGTA, which has high selectivity for calcium ions, and EDTA, which broadly chelates divalent metal ions. [5]

Test

EGTA and EDTA were each added to the precipitate solution, and the dissolution behavior was observed.

Learn

Since the precipitate dissolved in EDTA but scarcely dissolved in EGTA, it was inferred that the precipitate mainly consisted of magnesium phosphate. This was likely caused by an increase in the medium’s pH. The introduced noxE enzyme regenerates NAD⁺ through the following reaction by consuming oxygen and protons. Consequently, proton consumption likely led to an increase in pH.

However, in typical M9 medium cultivation with glucose as the carbon source, E. coli produces acidic compounds such as acetate through glycolysis, resulting in a tendency for pH to decrease. In this experiment, xylitol was used as the carbon source, which generates less acidic byproducts, thereby creating an environment prone to pH elevation.

To accurately determine xylitol growth efficiency in the future, it will be necessary to add a buffering system to suppress precipitate formation.

Furthermore, since the intestinal environment is maintained at a weakly acidic to neutral pH, such precipitation is unlikely to occur in vivo.

Cycle5

design

Since in Cycle 3 and Cycle 4 it was found that Xylego could grow in medium containing only xylitol, we decided to test whether it could survive when introduced into an environment containing other bacteria, simulating conditions within the gut.

Build

To reproduce an environment with other bacteria, co-culture was conducted using the strain harboring only the empty pTf16 vector (previously used as a negative control) and the strain expressing only the metabolic enzymes.

Test

The strain expressing only the metabolic enzymes was ampicillin-resistant (Amp^R), while pTf16 carried chloramphenicol resistance (Cm^R). After co-culturing the two strains for two nights, the cultures were serially diluted from 1× to 10⁷× and spread onto plates containing chloramphenicol and plates containing ampicillin, respectively. The appearance of colonies on both types of plates confirmed the coexistence of both strains. The number of colonies and the E. coli cell count per medium were as follows.

Additionally, Sample 5 showed no change in CFU count across all dilution series and differed greatly from the other two samples, so it was treated as an outlier. While pBl was barely detected at the 10⁷ dilution, pBl-xylB-xdh-noxE was detected at approximately 1.0×10¹⁰ CFU/mL. Compared with pTf16, the CFU count was about one-hundredth as high.。

Learn design

These results indicate that the strain capable of metabolizing xylitol could survive in xylitol-containing medium even in the presence of pre-established strains, without being completely outcompeted for nutrients.

In future experiments, we will co-culture with other probiotic species (not E. coli) to examine whether the xylitol-metabolizing strain can still persist under nutrient competition.

Cycle6

design

Measurement of Each Protein by SDS-PAGE

E. coli cells were cultured until OD600 reached 0.4, after which IPTG was added to a final concentration of 1 mM, and incubation was continued at 37 °C for 3 hours. The cells were then harvested by centrifugation at 6000×g for 10 minutes, washed with PBS, and resuspended in SDS-PAGE sample buffer containing BPB, DTT, and SDS. Samples were heated at 95 °C for 3 minutes before electrophoresis. A 17.5% acrylamide gel was used, and staining was performed with Sypro Ruby. Lane ① contained the LMW marker, with a total of 4 µg loaded. Lane ⑫ was E. coli carrying the negative control plasmid pTf16; lane ⑬ carried pBl-xylB-xdh-noxE; and lane ⑭ carried both pBl-xylB-xdh-noxE and pTf16-xytB-xytC-xytD-xytE.

Before electrophoresis, lanes ⑫ and ⑬ were concentrated 8-fold, and lane ⑭ was concentrated 120-fold.

Protein quantification was performed using Fiji and the Band Peak Quantification method [6]. When comparing to the marker, the bands at 97, 66, 20.1, and 14.4 kDa were used to determine the relationship between band signal intensity and protein amount.

Figure. Lane ① represents the Precision Plus Protein WesternC Standards. Lanes ①–⑪ are unrelated to this analysis. Lane ⑫ corresponds to pTf16, lane ⑬ to pBl-xylB-xdh-noxE, and lane ⑭ to pBl-xylB-xdh-noxE with pTf16-xytBCDE.

Table5. Protein expression levels in pBl-xylB-xdh-noxE measured by SDS-PAGE(The protein expression level for each bacterial cell was calculated as OD=1.0.)

Table6. Protein expression levels in pBl-xylB-xdh-noxE, pTf16-xytBCDE measured by SDS-PAGE(The protein expression level for each bacterial cell was calculated as OD=1.0.)

As a result, the strain containing only the xylitol metabolic enzyme cluster (pBl-xylB-xdh-noxE) expressed approximately 10 times more of these enzymes than the strain co-expressing both the metabolic enzymes and the transporter (pBl-xylB-xdh-noxE, pTf16-xytBCDE), confirming the superior efficiency of pBl-xylB-xdh-noxE in xylitol metabolism.

Furthermore, it was also confirmed that the strain with transporter introduction could not sufficiently express the transporter, indicating that it experienced a significant metabolic burden.

Mucosal healing

Cycle1

design

The desired characteristic of our recombinant E. coli is the ability to produce proteins that contribute to the treatment of IBD and secrete them extracellularly.

Build

As candidate proteins to be produced by E. coli, we considered not only EGF, which promotes the growth of epithelial cells, but also anti-inflammatory cytokines such as IL-10 and antibodies against inflammatory cytokines such as TNF-α.

There are two secretion mechanisms. The first is a two-step process in which the protein is transported from the inner membrane to the periplasm and then passes through the outer membrane. The second is a one-step process in which the protein passes through a tubular protein complex, allowing direct transport from the cytoplasm to the extracellular space [7].

Test

Through interviews, we selected EGF as the target protein. Promoting the recovery of damaged intestinal epithelial cells can help separate blood macrophages, which induce inflammation, from the intestinal environment that causes inflammation, thereby contributing to therapy.

For the transport system, we chose the more efficient one-step mechanism.

Learn

For the secretion system, we selected the one derived from Erwinia chrysanthemi, consisting of PrtD, PrtE, and PrtF.

Cycle2

design

The ultimate goal of the experiment is to validate our project using an intestinal model based on cultured cells. The intestinal model enables the study of interactions among intestinal epithelial cells, macrophages, and bacteria [8].

Since the target intestinal model uses C2BBe1 cells, a subline of Caco-2 cells, it was first necessary to confirm whether the growth of C2BBe1 cells is promoted by Epidermal Growth Factor (EGF).

Build

We planned an experiment to measure cell proliferation after introducing a scratch to confluent C2BBe1 cells on a culture plate.

Test

Confluent C2BBe1 cells on the plate were scratched with a p200 tip, and EGF was added to Dulbecco’s Modified Eagle’s Medium (DMEM) containing 10% Fetal Bovine Serum (FBS). The wound area was measured at 0, 24, 48, and (72) hours. Measurements were taken at three different locations.

Learn

As cell proliferation progressed, the wound boundaries became difficult to distinguish, leading to misaligned imaging fields. Consequently, the results were not accurate and did not meet our expectations. Therefore, we decided to use time-lapse imaging for future experiments.

Cycle3

design

The Lipase ABC transporter Recognition Domain 3 (LARD3) serves as a signal for the secretion system composed of PrtD, PrtE, and PrtF.

A fusion protein consisting of EGF and LARD3 with an N-terminal 6×His tag under the control of the T7 promoter was inserted into pBluescript II SK(-). The N-terminal 6×His sequence is a common tag that enables purification using affinity beads and detection with anti-6×His antibodies.

In addition, the lac promoter and the PrtD, PrtE, and PrtF transporter genes were inserted into pTf16.

Build

The ed plasmids, pBl-6×His-EGF-LARD3 and pTf16-lac-PrtD-PrtE-PrtF, were constructed using HiFi Assembly.

After several attempts, plasmid production in E. coli DH5α was successfully achieved. The plasmids were then extracted and transformed into E. coli BL21(DE3).

Test

Cloning was successful, and the expected electrophoresis results were obtained.

Learn

From the electrophoresis results, it was confirmed that both plasmids were successfully introduced into E. coli.

Cycle4

Design

Before conducting assays using C2BBe1 cells, it was necessary to confirm that our recombinant E. coli was capable of producing and secreting the 6×His-EGF-LARD3 fusion protein.

Build

We tested protein production and secretion using SDS-PAGE and Western blotting.

Test

Cell lysates were obtained by ultrasonic disruption. The molecular weight of pBl-6×His-EGF-LARD3 is 18.7 kDa. Electrophoresis was performed using a 15% acrylamide gel. The fusion protein was detected only in the supernatant of the lysate from E. coli harboring pBl-6×His-EGF-LARD3.

Using Band Peak Quantification, the concentration of 6×His-EGF-LARD3 in the culture supernatant was determined to be approximately 0.67–1.71 µg/mL.

Quantification of total protein using the 2D-Quant method suggested that ultrasonic disruption may have failed during sample preparation, and that the protein concentration in the culture supernatant was relatively low.

Learn

However, secretion of the 6×His-EGF-LARD3 fusion protein could not be confirmed in E. coli harboring both pTf16-lac-PrtD-PrtE-PrtF and pBl-6×His-EGF-LARD3.

It was suggested that competition between the lac promoter and the T7 promoter (via the lacUV5 promoter and IPTG induction) may have interfered with expression.

Cycle5

design

Based on the findings from Cycle 4, we designed a recombinant E. coli strain in which PrtD, PrtE, and PrtF are expressed under the control of the constitutive Anderson promoter J23118. By using this promoter, we expected an enhancement in secretion efficiency.

Build

The ed plasmid pTf16-J23118-PrtD-PrtE-PrtF was constructed using HiFi Assembly.

After several attempts, plasmid production in E. coli DH5α was successfully achieved. The plasmid was then extracted and co-transformed with pBl-6×His-EGF-LARD3 into E. coli BL21(DE3).

Figure19. Map of pTf16-J23118-PrtD-PrtE-PrtF

Test

Cloning was successful, and the expected electrophoresis results were obtained.

Learn

The electrophoresis results confirmed that both plasmids had been successfully introduced into E. coli.

Cycle6

design

We aimed to confirm the production and secretion of the 6×His-EGF-LARD3 fusion protein by SDS-PAGE and Western blotting.

Build

Since LB medium is unsuitable for samples intended for use with C2BBe1 cells, in addition to samples cultured in LB medium, we also prepared samples cultured in DMEM. Furthermore, to improve quantitativeness in SDS-PAGE analysis, Sypro Ruby was used for staining instead of CBB-R.

Test

In the Western blot, a band of approximately 20 kDa was visible in lane ②. This band was expected to correspond to the 6×His-EGF-LARD3 fusion protein. However, because similar bands were also detected in the negative controls (lanes ⑩ and ⑫), it could not be conclusively identified as 6×His-EGF-LARD3.

Learn

Since EGF contains six cysteine residues, it is possible that inclusion bodies were formed. Revising the induction conditions for IPTG and verifying plasmid sequence accuracy through sequencing may improve the results.

Cycle7

design

Based on the results from Cycle 2, we determined that it was necessary to analyze wound healing in C2BBe1 cells using time-lapse imaging.

Build

We conducted time-lapse imaging of C2BBe1 cell wound healing over 48 hours.

Test

A monolayer of C2BBe1 cells grown on a plate was scratched using a p200 pipette tip, and images were captured every 3 hours for a total of 48 hours.

The medium used was DMEM containing 10% FBS, supplemented with EGF at the desired concentration.

The resulting GIF images showed the progression of cell migration.

After imaging, wound areas were measured using the Wound Healing Size Tool to determine the area of cell proliferation. The cell-covered area increased most rapidly when EGF concentration was 50 ng/mL.

Learn

If the goal is to focus specifically on the effect of EGF on cell growth, it is necessary to culture the cells under serum-starved conditions using DMEM without 10% FBS. However, it was pointed out that this differs from the in vivo intestinal environment, where growth factors are distributed throughout the tissue.

In contrast, this experiment revealed that as the EGF concentration approached 50 ng/mL, wound healing accelerated, whereas at higher concentrations, cytotoxicity occurred and healing speed decreased.

Therefore, maintaining EGF concentration below approximately 50 ng/mL is expected to minimize adverse effects.

In the future, we plan to test this system using an intestinal model.

For more details, refer to Design, Experiment, and Result sections.

Kill Switch

Cycle1

design

After inflammation is treated, if Xylego continues to survive, there is a risk of carcinogenesis due to excessive production of EGF [9].

To prevent this problem, we designed a system in which the bacteria undergo programmed cell death once inflammation has subsided.

Build

We focused on the toxin–antitoxin (TA) system as a potential mechanism for self-destruction in genetically engineered bacteria.

In this system, both toxin and antitoxin are constitutively expressed; however, under stress conditions that inhibit expression, the unstable antitoxin is degraded first, allowing the toxin to act and cause cell death [10].

In our , we aimed to use nitric oxide (NO)—one of the biomarkers of inflammatory bowel disease (IBD)—as the stress signal.

Since this system is widely conserved among many bacterial species, it was considered feasible to implement.

To evaluate the validity of this approach, we conducted an interview with an expert in the field.

Test

We interviewed a microbiology researcher about our proposed system. (For details, see Integrated Human Practice.))

Learn

It was found that this system is indeed applicable.

Furthermore, by employing restriction–modification (RM) systems [11], in which the toxin and antitoxin correspond to a restriction enzyme and its methyltransferase, respectively, it may be possible to induce genomic or plasmid DNA cleavage and thereby prevent the horizontal transfer of antibiotic resistance genes.

Cycle2

design

Because intestinal NO levels can fluctuate depending on diet, we decided instead to use tetrathionate as the signal marker.

Based on the findings from Cycle 1, we selected BsaI, a restriction enzyme that cleaves the ampicillin-resistance gene, for use in the kill-switch mechanism.

To enable E. coli to sense tetrathionate, we introduced the TtrSR two-component system.

Build

Using HiFi Assembly, we constructed the following plasmids and introduced them into E. coli BL21(DE3).

Figure2. Map of pBluescriptⅡSK(-)-TtrR-BsaIM1-BsaIM2-BsaIR

Test

Cloning was successful, and the expected electrophoresis results were obtained.

Learn

The electrophoresis results confirmed that both plasmids were successfully introduced into E. coli.

Next, we plan to verify whether TtrS—the tetrathionate sensor protein—is expressed on the cell surface and whether the kill-switch functions properly.

Cycle3

design

To evaluate the function of the kill-switch, it was necessary to determine whether the system can respond to tetrathionate.

Therefore, we designed an experiment to test this.

Build

We planned to culture E. coli in a medium containing tetrathionate (representing the inflamed condition) and then transfer it to a medium without tetrathionate (post-inflammation).

If the kill-switch functions correctly, cell death should occur upon tetrathionate depletion.

We used sodium tetrathionate as the source of tetrathionate for this experiment.

Test

E. coli cultured in liquid medium containing 1 mM tetrathionate was transferred to plates and liquid media with different tetrathionate concentrations (1, 0.5, 0.25, 0.125, and 0 mM).

Colony numbers on the plates were counted after incubation, and OD measurements in the liquid cultures were taken every 30 minutes.

Results after overnight incubation on plates containing 1, 0.5, 0.25, 0.125, and 0 mM tetrathionate.

Learn

Colonies were observed at all tetrathionate concentrations, and growth was also detected in liquid cultures.

These results indicate that the kill-switch did not function properly, and that the bacteria survived even under tetrathionate-depleted (post-treatment) conditions.

Cycle4

design

Based on the results from Cycle 3, we concluded that it was necessary to confirm whether TtrSR was being expressed and functioning as a tetrathionate-sensing system.

To do this, we designed an experiment to visualize gene expression via GFP fluorescence under tetrathionate induction.

Build

To enable visualization by GFP, we constructed new plasmids and introduced them along with pTf16-TtrS into E. coli.

Test

The recombinant E. coli strains were inoculated into media with and without tetrathionate and cultured overnight. Fluorescence expression was then observed.

Learn

Since fluorescence was observed in the tetrathionate-containing culture, we confirmed that the bacteria were able to sense tetrathionate.

However, fluorescence was also detected in the culture without tetrathionate, suggesting promoter leakage.

A possible explanation is that TtrS—which phosphorylates TtrR—was overexpressed, leading to constitutive phosphorylation of TtrR even in the absence of tetrathionate.

Cycle5

design

From the results of Cycle 4, it was confirmed that this E. coli strain can sense tetrathionate. Therefore, the next step was to verify whether the restriction enzyme could cleave the plasmid and genomic DNA of this E. coli, thereby inducing cell death [12].

Build

To conduct this verification, we newly synthesized the following plasmid.

Test

The E. coli strain harboring only pBl-BsaIR, in which expression was pre-induced for 3 hours with IPTG, was cultured in liquid medium, serially diluted, spread onto plates, and incubated overnight.

Colony-forming units (CFU) were then measured.

Learn

As shown in the photo, a lawn of colonies grew over the entire surface, so CFU could not be measured. Moreover, even on the dilution plates, every plate developed a lawn, indicating that neither the plasmid nor the genome was cleaved by BsaIR. From this experiment, we concluded that either the enzymes were not expressed properly or they lacked activity.

Cycle 6

Design

While investigating the reasons for the kill-switch failure, we found that the RBS used for the methyltransferases Bsa1M1, Bsa1M2, and the restriction enzyme Bsa1R were all BBa_B0034. Consequently, the expression levels of the restriction enzyme (Bsa1R) and the methyltransferases (Bsa1M1, M2) were the same. We realized this could cause the Kill Switch to fire even when tetrathionate concentrations had not yet fallen sufficiently, due to high-level expression of the restriction enzyme. We will test this hypothesis with a model.

Build

We will evaluate a model of the Kill Switch under two conditions: (1) Bsa1R and Bsa1M1/M2 expressed at the same level, and (2) Bsa1R expressed at 0.3× the level of Bsa1M1/M2.

Test

• When Bsa1R expression is set to 0.3× that of Bsa1M1/Bsa1M2, the Kill Switch is triggered 2 hours later than when Bsa1R is expressed at the same level as Bsa1M1/Bsa1M2.
• When Bsa1R expression equals that of Bsa1M1/Bsa1M2, the Kill Switch triggers at a promoter leak level of 40% of the nominal value.
• When Bsa1R expression is 0.3× that of Bsa1M1/Bsa1M2, the Kill Switch triggers at a promoter leak level of 20% of the nominal value.

Learn

It is considered that reducing the expression level of the restriction enzyme to about 0.3× that of the methyltransferases can decrease the likelihood of the Kill Switch being falsely triggered when the tetrathionate level remains high. Although the precise relationship between tetrathionate concentration and the activity of the tetrathionate-inducible promoter still needs to be investigated, we decided to use BBa_B0032 as the ribosome binding site for Bsa1R to reduce its expression level.