Engineering Success
Our project, PreserVEG, aims to reduce food waste by improving produce safety and enabling the visualization of deterioration. To accomplish this aim, we used Cre/loxP-based three-in-one system. The experiments to realize this system can be divided into three main elements. The first project involved the introduction and functional verification of the Cre plasmid, the second focused on the construction of the loxP plasmid and functional validation of nisin, and the third examined the expression and functional simulation of TOM. These three projects are interconnected as follows: Co-transformation of the Cre and loxP plasmids Co-culture simulation of E. coli expressing TOM with E. coli expressing EtnR1 and EtnR2 Based on these two relationships, the three projects proceeded in parallel, through repeated cycles of experimentation and simulation, to improve the overall performance of the system.
To induce Cre recombinase expression in response to ethylene, we decided to utilize the ethylene metabolic control system from mycobacterium chubuence NBB4. Specifically, we applied the transcription factors EtnR1 and EtnR2, which are activated by epoxyethane, an ethylene-derived metabolite. We aimed to construct a base plasmid for protein expression using epoxyethane as an inducer by introducing Petn, which binds to these transcription factors. (*1)
We performed GoldenGate Assembly to construct a regulatory plasmid expressing the transcription factor EtnR1R2, which is involved in ethylene metabolism. Whole plasmid sequencing revealed a single nucleotide variation compared to the expected sequence, but this did not affect the insert (EtnR1R2) or plasmid replication.
Thus, we successfully constructed a plasmid containing EtnR1R2.
We constructed a plasmid equipped with a complete regulatory system that expresses the transcription factors EtnR1R2, activated by the ethylene metabolite epoxyethane, and can subsequently be connected to the Petn promoter control system.
Using the replication origin (BBa_J435300) and lac promoter (BBa_J435350) included in the iGEM 2025 Distribution Kit as a base, the region containing the BsaI recognition sequence was amplified by PCR. Additionally, fragments of the transcription factor EtnR1R2, essential for ethylene sensing, were synthesized by IDT. These three fragments were inserted using the Golden Gate Assembly (using BsaI restriction).
Plasmid was extracted from colonies obtained via transformation and designated as the LacP_EtnR1R2_GoldenGate plasmid. Since all sequences used in this assembly were amplified by PCR, full-length sequencing analysis was carried out to confirm the absence of mutations.
Alignment with the expected sequence revealed a single-base mutation at the origin of replication.
Although the constructed plasmid contained a single-base mutation compared to the expected sequence, the transformants proliferated normally in liquid medium under ampicillin selection. This suggests the mutation does not significantly affect E. coli growth function. Therefore, we concluded that we had successfully constructed the plasmid containing EtnR1R2, which forms the basis for ethylene metabolite sensing, and proceeded to the next step.
Using the plasmid constructed in Cycle1-1 as the backbone, Gibson Assembly was performed with Petn_fuGFP as the insert. Whole plasmid sequencing confirmed the constructed plasmid perfectly matched the intended sequence, successfully establishing a plasmid based on EtnR1R2 and Petn.
We constructed a plasmid equipped with a complete set of regulators to enable protein expression using epoxyethane as an inducer, incorporating Petn activated by EtnR1R2. Furthermore, to enable subsequent functional analysis of EtnR1R2 and Petn, a simple detectable GFP was introduced downstream of Petn.
Using the LacP_EtnR1R2_GoldenGate plasmid constructed in Cycle1-1 as the backbone, Petn_fuGFP was introduced via Gibson Assembly.
Plasmid was extracted from colonies obtained via transformation, resulting in the EtnR1R2_Petn_fuGFP_Gibson plasmid. Since all sequences used in this assembly were amplified by PCR, full-length sequencing was performed to confirm the absence of mutations.
The results confirmed that the constructed plasmid possesses the expected nucleotide sequence.
We successfully constructed a plasmid containing Petn, which is regulated by transcription factors EtnR1 and EtnR2, along with GFP necessary for functional analysis of these factors.
Here, we aimed to construct a circuit that activates EtnR1/EtnR2 using epoxyethane, an ethylene monoxide, to induce GFP expression. However, fluorescence could not be detected even after optimizing the concentration conditions and induction timing of inducers such as IPTG and epoxyethane (hereafter EtO). SDS-PAGE results revealed that EtnR1 was found in the soluble fraction, while EtnR2 was present in the insoluble fraction. Therefore, it was suggested that the reason why the induction circuit by epoxyethane did not work was due to the folding failure of EtnR2. It became clear that improvements such as codon optimization are necessary in the future.
After inducing EtnR1/EtnR2 expression with IPTG, we aimed to activate epoxyethane and induce GFP expression. GFP fluorescence was monitored by adding epoxyethane at concentrations ranging from 0 to 10 µM. However, GFP fluorescence was not detected under any of these conditions. Since bacterial growth was unaffected, the absence of GFP fluorescence was attributed to insufficient epoxyethane for induction, rather than toxicity.
EtnR1 and EtnR2 are transcription factors that bind to the Petn promoter region in the presence of epoxyethane and induce transcription of downstream genes. Therefore, we thought that placing GFP downstream of Petn and confirming its expression would allow simultaneous verification of EtnR1/R2 expression and function, as well as the effectiveness of Petn as a promoter.
Epoxyethane (500 µg/mL in DMSO) was added to the culture medium at concentrations of 0/1/5/10 μM, and IPTG was added to achieve a concentration of 0.4 mM. The culture was incubated overnight at 25°C. Samples were then collected, and GFP fluorescence was confirmed under black light illumination. Note that epoxyethane may exhibit cytotoxicity at high concentrations (*2), so concentrations were set considering its potential impact on E. coli growth inhibition.
GFP fluorescence was not detected at any concentration condition. Furthermore, no difference in E. coli growth was observed after protein expression induction by IPTG and epoxyethane.
Although epoxyethane is a highly toxic substance raising concerns about cell growth inhibition, no correlation was observed between epoxyethane concentration and OD600 values after induction. Therefore, it was confirmed that E. coli growth is not inhibited within the concentration range used in this experiment. This led us to consider the possibility that the epoxyethane concentration had not reached the threshold for EtnR1R2 activation. Consequently, we decided to attempt expression induction under conditions with a higher concentration of epoxyethane.
To activate EtnR1/EtnR2 and induce GFP expression using epoxyethane, we conducted verification using higher concentrations of epoxyethane (0-50 µM) than in the previous cycle. After adding 0.4 µM IPTG and culturing overnight, fluorescence was checked, but no GFP fluorescence was observed under any conditions.
In Cycle 2-1, epoxyethane was added at final concentrations of 0-10 µM, but GFP fluorescence was not detected. Therefore, we decided to set the epoxyethane concentration even higher and re-verify whether the EtnR1/EtnR2 system responds.
Previous studies reported that in the EtnR1R2 and Petn expression regulation system, fluorescence values peaked (*1) when 50 mM epoxyethane was added. Therefore, in this cycle, epoxyethane was added to the culture medium to achieve final concentrations of 0/1/5/10/25/50 µM. Additionally, IPTG was added to achieve a concentration of 0.4 mM, and the culture was incubated overnight at 25°C. After incubation, samples were collected, cells were harvested, and GFP fluorescence was observed.
Under black light illumination, no GFP fluorescence was detected at any concentration condition. Sample numbers 0 through 6 correspond to epoxyethane final concentrations of 0/1/5/10/25/50 µM, respectively.
The failure to detect GFP expression even at high epoxyethane concentrations suggests that insufficient epoxyethane concentration alone may not be the sole cause of the lack of GFP expression. Previous studies have reported that pre-induction of EtnR1/R2 by IPTG enhances fluorescence intensity (*1). This led us to consider that differences in the expression state of transcription factors might influence activation efficiency.
After induction of the transcription factors EtnR1 and EtnR2, epoxyethane was added to examine whether activation of these transcription factors would lead to a GFP fluorescence response. After culturing for 8 h or 16 h with 0.2 or 0.4 mM IPTG added, epoxyethane was added. However, GFP fluorescence was not detected under either condition. Based on these results, concerns arise that the induction response to EtO is not established, or that there may be issues with the expression of EtnR1 and EtnR2 itself. Therefore, in the next cycle, we decided to first examine the expression of EtnR1 and EtnR2.
Previous reports indicate that pre-inducing EtnR1/R2 before adding epoxyethane enhances the GFP fluorescence response (*1). Furthermore, epoxyethane is highly reactive with water, raising the possibility that it decomposes before IPTG-induced EtnR1/R2 expression is complete, thereby losing its function as an inducer. Therefore, in this cycle, we attempted to improve the fluorescence response by adding epoxyethane after pre-inducing the transcription factors with IPTG.
IPTG was added at concentrations of 0.2 or 0.4 mM, and after incubation for 8 or 16 hours, epoxyethane was added. The simultaneous addition of IPTG and epoxyethane—that is, the concurrent induction of the transcription factors and GFP under the Petn promoter—was used as a control for comparison.
In Cycle 2-2, it was confirmed that the addition of epoxyethane at a final concentration of 50 mM did not inhibit the growth of E. coli. Therefore, the same concentration was adopted in this experiment.After incubation at 25 °C, samples were collected, cells were harvested, and GFP fluorescence was observed.
Under black light illumination, GFP fluorescence was not detected at any concentration condition. Refer to the Build table for sample numbers.
We examined multiple conditions, including pre-induction with IPTG and simultaneous induction, but GFP fluorescence was not observed in any case. This suggests that either the induction response to epoxyethane is not occurring at all, or EtnR1/EtnR2 itself is not being expressed. Therefore, we decided to seek advice from experts and perform troubleshooting.
A meeting with Prof. Yokokawa suggested that EtnR1/EtnR2 were not expressing correctly, indicating the Petn expression system might not be functioning. Originally, the Gifu team's wet lab lacked the necessary reagents and equipment to perform SDS, but thanks to Prof. Yokokawa's generosity, SDS-PAGE was conducted.
SDS analysis confirmed the presence of an EtnR1 band in the soluble fraction, while an EtnR2 band was observed in the insoluble fraction. Based on these results and the fact that the sequence has not been codon optimized, it is possible that the induction circuit by epoxyethane did not function due to faulty folding of EtnR2.
We consulted Professor Yokokawa to determine why GFP fluorescence was not observed in our prior evaluations and to refine the experimental setup. As a result, it was pointed out that the transcription factors EtnR1 and EtnR2, derived from different microorganisms, might not be expressing properly. Therefore, with the professor's cooperation, we decided to use SDS-PAGE to confirm whether the EtnR1/EtnR2 proteins were expressing in E. coli.
cre detail about Human Practices, click here
The Gifu team's wet lab lacked the necessary reagents and equipment, making SDS-PAGE difficult to perform. However, thanks to Professor Yokokawa's generosity, we were able to conduct the experiment using his original protocol. For detailed experimental methods, please see Experiment.
E. coli cells induced only with IPTG (0.4 mM) were lysed, and both soluble and insoluble fractions were applied. Previous studies have reported bands for EtnR1 around 50-75 kDa and for EtnR2 around 25 kDa (*3). In the present results, a band for EtnR1 was detected in the soluble fraction at around 66 kDa, whereas a strong band for EtnR2 was observed in the insoluble fraction at around 25 kDa. These results suggest that although EtnR2 is expressed, it may not be functioning properly due to misfolding.
SDS-PAGE results confirmed the expression of EtnR1 and EtnR2. However, EtnR2 was present in the insoluble fraction and was likely nonfunctional due to folding defects. This suggested that interaction between the two transcription factors was not established, and the epoxyethane-induced expression system may not have functioned.
Furthermore, since the EtnR1/EtnR2 sequences were not codon optimized for E. coli, there was concern that this might cause reduced translation efficiency and folding defects.
However, due to time constraints, it was difficult to synthesize new codon-optimized genes and redo all the assays. Therefore, we decided to try to improve the expression status using the Rosetta2 strain, which possesses tRNA corresponding to rare codons.
In this cycle, we constructed a Cre expression plasmid with the goal of “detecting ethylene metabolites to express Cre and activate the Cre/LoxP system.” First, we replaced the GFP sequence in the EtnR1R2_Petn_fuGFP_Gibson plasmid used in Cycle 2 with Cre, creating the EtnR1R2_Petn_Cre_Gibson plasmid.
This plasmid was transferred to the Nagahama team. To confirm that EtnR1 and EtnR2 were expressed and transferred to the soluble fraction, the host strain was changed to Rosetta2 and protein expression analysis was performed. SDS-PAGE results showed no bands corresponding to EtnR1 and EtnR2, indicating that further investigation of expression levels and folding conditions is necessary.
A plasmid containing Cre downstream of EtnR1/EtnR2 and Petn was constructed to induce Cre expression using epoxyethane as an inducer. DNA sequencing of the newly introduced Cre confirmed perfect match with the expected sequence. Thus, successful construction of the plasmid containing EtnR1R2 was achieved.
We aimed to construct a plasmid in which Cre expression is induced by ethylene metabolites. The functionality of the newly introduced Cre (BBa_) via Golden Gate Assembly was characterized in LoxP Cycle 3. This Cre recombinase recognizes loxP sites and mediates excision of the loxP-flanked sequence.
The EtnR1R2_Petn_fuGFP_Gibson plasmid constructed in Cycle 1-2 contained BsaI sites flanking the GFP cassette. We synthesized a Cre DNA fragment bearing compatible BsaI sites and, via Golden Gate Assembly, replaced the GFP cassette with the Cre sequence.
Following transformation, plasmid DNA was extracted from resulting colonies and designated the EtnR1R2_Petn_Cre_GoldenGate plasmid. Partial sequencing across the Cre insert confirmed that the fragment matched the intended sequence.
Thus, we successfully constructed a plasmid encoding EtnR1 and EtnR2, the Petn promoter they regulate, and Cre recombinase under Petn control. The completed plasmid was transferred to the Nagahama wet lab.
Taking into account the Cycle 2 results and the lack of codon optimization, we attempted to improve EtnR1/EtnR2 expression by using the Rosetta2 strain, which supplies tRNAs for rare codons. Rosetta2 was transformed with the EtnR1R2_Petn_Cre_GoldenGate plasmid constructed in Cycle 3-1. We then performed SDS-PAGE on IPTG-induced samples; however, no bands corresponding to either transcription factor were detected.
Building on Cycle 2, where EtnR2 appeared insoluble in BL21, we switched to the Rosetta2 strain to supplement rare-codon tRNAs and reassess expression. Rosetta2 cells transformed with the EtnR1R2_Petn_Cre_GoldenGate plasmid were processed, and protein samples were concentrated using centrifugal filter units: 12,000 rpm for 30 min, then the filter insert was inverted and spun at 3,500 rpm for 2 min to recover the retentate.
The concentrated samples were separated into supernatant (soluble) and pellet (insoluble) fractions and analyzed by SDS–PAGE under both IPTG-induced and non-induced conditions.
E. coli subjected only to IPTG induction (0.1 mM) was lysed, and both soluble and insoluble fractions were applied. Previous studies have identified bands for EtnR1 around 50–75 kDa and for EtnR2 around 25 kDa (*3). In this study, no bands for EtnR1 or EtnR2 were detected.
The reason EtnR1R2 expression was confirmed at Gifu University but not at Nagahama Institute of Bio-Science and Technology may be due to differences in culture conditions. At Gifu University, the culture was incubated at 37°C for 3 hours during this experiment, whereas at Nagahama Institute of Bio-Science and Technology, incubation was performed at 37°C for approximately 22 hours. As a result, the protein may have denatured, preventing the detection of bands at the intended positions.
We created a loxP plasmid susceptible to Cre recombinase. In this project, we designed the loxP sequences to be oriented in the same direction, enabling the gene between the loxP sites to be knocked out via deletion. By placing the antimicrobial peptide Nisin and a terminator between the loxP sites and introducing GFP downstream of the loxP, we aimed to construct a plasmid capable of two-step gene expression in a Cre-dependent manner.
We designed and constructed a plasmid to enable the sequential expression of the antibacterial peptide NisinQ and the fluorescent protein GFP using the Cre/loxP system. The plasmid, constructed via Gibson Assembly, was confirmed by sequence analysis, and we successfully built a vector containing NisinQ between loxP sites.
To achieve both antibacterial and visualization functions within a single metabolic pathway, it was necessary to design a plasmid enabling sequential expression of the antibacterial peptide NisinQ and the fluorescent protein GFP.
To accomplish this, we applied the Cre/loxP system, combining conditional knockout with Cre expression-dependent gene regulation.
The sequences for loxP, terminator, and GFP were quoted from a paper based on an E. coli assay where the terminator was inserted between loxP sites to halt expression, allowing downstream GFP expression after recombination (*1). Furthermore, considering molecular weight and solubility in E. coli, the composite part (BBa_K4437002) designed for iGEM Calgary 2022 (*2) combining NisinQ with nusA was adopted.
Gibson Assembly was performed using the downstream loxP sequence and the sequence containing GFP as the vector, and the upstream loxP sequence, NusA_NisinQ, and the terminator sequence as the insert.
Plasmid was extracted using colonies obtained through transformation and designated as the loxP_NisinQ_sfGFP_Vector_Assembled plasmid. Since all sequences used in this assembly were amplified by PCR, full-length sequence analysis was performed to confirm the absence of mutations.
As a result, it was confirmed that the plasmid contained the designed sequence.
We successfully constructed a plasmid containing the antibacterial peptide NisinQ between loxP sites.
In the next step, we will verify whether the loxP sites are actually knocked out by Cre recombinase.
Site-specific cleavage using the Cre/loxP system is an essential process for advancing this project. Therefore, we conducted an in vitro recombination assay using purified Cre recombinase.
First, Cre recombinase and the loxP plasmid were reacted under NEB-recommended conditions. Subsequently, the loxP plasmid was linearized using restriction enzymes to investigate the relationship between plasmid form and enzyme reaction. Furthermore, conditions for enzyme reaction time and enzyme concentration were optimized. The results of these verifications were evaluated by confirming the band of the loxP-deleted plasmid via electrophoresis.
Based on the NEB recommended protocol, an in vitro recombination assay of loxP plasmids using purified Cre recombinase was performed. The positive control loxP 2+ yielded the recombinant fragment. In contrast, no new bands resulting from recombination were detected in the plasmid constructed in Cycle 1. These results suggest that constraints arising from the three-dimensional structure due to supercoiling, or sequences surrounding the loxP site, may be inhibiting Cre access
To verify whether the constructed loxP plasmid is cleaved by Cre recombinase, we decided to use purified Cre recombinase (NEB#M0298). First, Cre and the loxP plasmid were reacted under NEB's recommended conditions (*3). Subsequently, reaction conditions would be optimized as necessary. The success criteria for the assay were detecting the predicted fragment lengths [3006 kb, 1922 kb] by electrophoresis and no new bands detected in the negative control.
Using the loxP plasmid purified by mini-prep in Cycle 1 as a sample, we set up a reaction system employing purified Cre recombinase and analyzed the plasmid strand length in the post-treatment sample via agarose gel electrophoresis. To ensure verification reliability, we also assayed a LoxP2+ Cre+ sample (positive control) and a loxP plasmid Cre- sample (negative control) under identical conditions.
Following the NEB protocol, a recombination assay was performed using purified Cre recombinase, and the strand lengths of the treated plasmids were confirmed by electrophoresis.
From the official NEB website(*3), pLox2+ is known to be a circular plasmid of 2787 bp (appears to be around 1.7 kbp) and a fragment of 838 bp after recombination. In this experiment, no band was observed for the positive control fragment. However, a band likely representing a circular plasmid was detected around 1.7 kbp, indicating that the positive control recombination was successful. On the other hand, the plasmid we constructed did not change the band pattern with or without Cre, suggesting that recombination failed.
Under these conditions, recombination succeeded in the positive control, but no recombination was observed in the constructed loxP plasmid. This result suggests that the supercoiled structure may restrict Cre recombinase access to the loxP site. In the next cycle, it will be necessary to investigate differences in recombination efficiency due to plasmid topology.
To evaluate the DNA structure-dependent reactivity of the loxP plasmid, the plasmid was linearized with SpeI to enhance the accessibility of purified Cre recombinase near the loxP site, followed by verification. Electrophoresis results showed no new bands from cleavage in the circular (supercoiled) form, while recombinant products were detected in the linearized plasmid. This suggests that the DNA's three-dimensional structure affects the reaction efficiency of Cre recombinase.
Based on Human Practice with Prof. Yokogawa and literature review, we hypothesized that in circular plasmids, the supercoiled structure may reduce Cre recombinase access to loxP sites, potentially lowering recombination efficiency. Furthermore, based on the random collision model hypothesis (*4), we considered the possibility that two rings generated by Cre cleavage could become entangled, preventing detection of the recombinant plasmid at the correct size.
Therefore, we designed the plasmid to be processed with the restriction enzyme SpeI to linearize it and expose the region near loxP, enabling Cre recombinase to act efficiently.
For more detail about Human Practices, click here
After mini prep, the loxP plasmid (4928 bp) was digested with SpeI to prepare linearized plasmid.
After the reaction, the enzyme was inactivated, and the product was purified using a gel extraction kit. The linearized plasmid obtained from purification was used in the reaction with Cre recombinase. The same experiment was performed with a SpeI-untreated sample. This allowed comparison of recombination efficiency based on topological differences.
SpeI does not cleave the loxP sequence or the reporter region; therefore, it was determined that it does not affect the DNA strand length resulting after recombination.
Analysis of reaction products by electrophoresis revealed that no bands indicative of Cre-mediated recombination was detected in the circular loxP plasmid, just as in Cycle2-1. In contrast, the linearized loxP plasmid showed bands indicating Cre-mediated recombination, confirming the deletion between the loxp plasmids.
When recombination occurs in the constructed plasmid, a 1922 bp sequence is deleted between loxP sites, resulting in a 3006 bp plasmid. The band observed around 3000 bp in the linear Cre+ sample in Fig. is considered to be the recombinant plasmid. Furthermore, the band observed slightly above the arrow in the linear Cre- sample is thought to be either a recircularized plasmid or a restriction enzyme processing remnant.
It was revealed that linearizing the plasmid with restriction enzymes and resolving its supercoiled structure enables recombination at loxP sites by purified Cre recombinase. This supports the possibility that access to loxP is restricted in the supercoiled state due to steric hindrance. Furthermore, these results confirmed that the constructed plasmid functions as a recognition site for Cre recombinase.
However, Cre recombination in E. coli occurs with the plasmid in its circular form. Therefore, we have decided to investigate conditions further to achieve successful recombination in the circular plasmid.
To evaluate cleavage efficiency in circular (supercoiled) plasmids, we performed in vitro verification under all 20 conditions combining Cre recombinase enzyme amount and reaction time. Electrophoresis results revealed a novel fragment generated by loxP deletion under the high-concentration, short-incubation-time condition of 4 units of Cre recombinase and 5 and 15 minutes of reaction time. This confirmed that loxP site cleavage occurs even in the supercoiled state.
In the Cre/loxP system, increasing enzyme concentration and reaction time may improve recombination efficiency (*5). Therefore, to confirm Cre recombinase cleavage in circular plasmids, we set multiple conditions for Cre recombinase concentration and incubation time and verified the results.
Cre-recombined concentrations were set at four conditions: 1 unit, 2 units, 3 units, and 4 units. Each Cre concentration was incubated for 5 min, 15 min, 30 min, 1 h, and 2 h, resulting in a total of 20 conditions examined. Plasmid DNA was extracted from each condition using mini-prep and analyzed by electrophoresis.
Electrophoresis results showed bands likely representing recombinant plasmids when incubated for short periods.
Bands were particularly clearly visible in samples incubated for 5 minutes and 15 minutes after adding 4 units of Cre.
Increasing the Cre recombinase concentration and shortening the incubation time revealed that recombination by Cre recombinase occurs even in the supercoiled state.
Then, we decided to examine the Cre recombination assay within E. coli itself.
Based on discussions with Prof. Yokokawa and literature review, it was suggested that recombination mediated by Cre recombinase may occur more readily in vivo than in vitro (*6). Therefore, in this cycle, we performed in vivo validation of the Cre/loxP system using a Cre-expressing plasmid and the loxP plasmid constructed in Cycle 1. The cell culture after protein expression induction was used to examine whether recombination occurred at the DNA level. In addition, we quantified the GFP fluorescence intensity observed during bacterial harvesting to verify the progression of GFP protein expression after recombination. Furthermore, we aimed to comprehensively evaluate the effects of promoter leak expression and IPTG concentration on GFP expression levels by setting multiple induction conditions.
For more detail about Human Practices, click here
BL21 cells were co-transformed using a plasmid expressing Cre (BBa_25Q0IEK8) and a loxP plasmid. After transformation, colonies were obtained and the strains required for the in vivo assay were successfully generated. To confirm recombination success, IPTG induction was performed to visually detect GFP expression; however, no fluorescence was observed in any sample. Therefore, considering the possibility that Cre recombinase did not cleave the loxP site, we decided to proceed with DNA-level analysis as the next step.
To verify recombination by Cre/loxP system in E. coli, Cre plasmid and loxP plasmid were co-transformed into E. coli BL21(DE3). Using this strain, it is possible to verify recombination at the DNA level and to examine protein expression from the loxP plasmid after recombination.
Co-transformation (details in experiment) was performed (*7). To prevent incompatibility, Cre and loxP plasmids were selected with different replication origins. The former carries an ampicillin resistance gene, while the latter carries a chloramphenicol resistance gene. Culturing on medium containing both ampicillin and chloramphenicol enabled reliable screening of colonies successful in co-transformation.
Transformation succeeded, and colonies were obtained. A glycerol stock was prepared.
We successfully created E. coli strains harboring the recombinase expression plasmid and a plasmid containing the loxP sequence, which is the recognition site for the enzyme, both necessary for verifying the Cre/loxP system. We decided to use this strain to perform an in vivo Cre recombination assay.
To confirm recombination occurs in E. coli via the Cre/loxP system, we used the strain created in Cycle3-1-1. We aimed to induce Cre expression in this bacterial culture and delete the sequence between the loxP sites. As in Cycle 1, the Cre-sensitive plasmid contains a terminator between the loxP sites and a GFP-coding sequence downstream of the loxP sites. Therefore, confirmation of GFP expression following the deletion of the sequence between the loxP sites allows verification of whether Cre recombination was successful.
E. coli harboring Cre and loxP plasmids were cultured overnight at 25°C with IPTG added at three concentrations: 0, 0.2, and 0.4 mM. After incubation, GFP expression was confirmed under each condition.
Under any conditions, GFP fluorescence could not be detected in liquid medium. This may be due to the Cre recombinase failing to cleave the loxP site. Therefore, we decided to verify whether the loxP site was cleaved in the next cycle.
A possible reason for the lack of GFP fluorescence is that Cre recombinase did not cleave at the loxP sites. Therefore, in the next cycle, we will verify whether cleavage at the loxP sites has occurred.
To verify whether site-specific cleavage by the Cre/loxP system functions in E. coli, experiments were conducted using E. coli prepared with Cycle3-1. Plasmids were extracted from samples after IPTG induction and subjected to electrophoresis to confirm the activity of Cre recombinase in vivo. The results showed recombination was confirmed under all conditions, including samples without IPTG addition.
Using E. coli containing the Cre plasmid and loxP plasmid created in Cycle 3-1, verify recombination at the DNA level. The cause of the lack of GFP fluorescence expression is thought to be one of the following three possibilities.
Cause 1 → Recombination by Cre did not occur, and the polymerase did not reach the GFP gene.
Cause 2 → Recombination by Cre occurred, but the protein was not expressed.
Cause 3 → Recombination by Cre occurred, and the GFP protein was expressed, but it lacks activity.
The goal was to narrow down the cause(s) preventing fluorescence observation to one, or two or three possibilities, by confirming the deletion at the DNA level.
E. coli harboring Cre and loxP plasmids were cultured overnight at 25°C with IPTG added at four concentrations: 0, 0.1, 0.2, and 0.4 mM. Plasmids were recovered from post-induction samples via mini prep. These extracted plasmids were subjected to electrophoresis. Recombination was verified by confirming the presence of bands corresponding to plasmids with the loxP region deleted.
We unexpectedly observed GFP fluorescence when recovering the bacterial cells during the initial stage of miniprep.
Furthermore, we confirmed the band of loxP plasmid after recombination in all samples. The position of this band was consistent with the results of the in vitro assay. This result indicates the success of the in vivo Cre recombination assay. The detection of bands in the IPTG (-) samples suggests that Cre leakage expression occurred.
*loxP plasmid = 4928 bp(mini prep), Cre colonal plasmid = 3353 bp (synthetic nucleic acid) We applied each sample to the wells in equal amounts of DNA.
As GFP fluorescence was confirmed during bacterial cell recovery, all causes indicated in the Design phase could be rejected.
And the electrophoresis results demonstrated that the Cre/LoxP system functions in co-transformed E. coli. Even in IPTG (-) samples, bands following recombination were clearly visible. Therefore, even in the absence of IPTG, Cre can express in sufficient quantities due to leakage from the T7 promoter.
However, in the IPTG (-) sample in Fig., we didn't observe fluorescence visually, which suggests that the T7 promoter in the loxP plasmid can highly regulate protein expression.
In the next step, it is necessary to quantify the GFP fluorescence and compare the fluorescence intensity for each IPTG concentration.
GFP fluorescence intensity was measured under each induction condition (IPTG 0/0.1/0.2/0.4 mM), and the relative values were compared. Based on these results, the effects of Cre leak expression and IPTG concentration on GFP expression levels were examined. After protein expression induction, the culture medium OD600 was adjusted to 0.5, and fluorescence values were measured. The results showed maximum fluorescence at 0.1 mM IPTG, with fluorescence decreasing at higher concentrations. While the confirmation of recombination at the DNA level suggested the influence of Cre leak expression, a significant difference was confirmed between the fluorescence values of samples without IPTG addition and those with IPTG addition.
Confirmation of GFP expression suggests the success of the Cre recombination and the proper progression of the gene circuit in the loxP plasmid after recombination.
We aim to compare GFP fluorescence intensity with and without IPTG and confirm that fluorescence values are significantly higher in the presence of IPTG. Additionally, we will investigate the extent to which Cre leak expression and the resulting recombination affect gene expression from the loxP plasmid.
E. coli harboring Cre and loxP plasmids were cultured overnight at 25°C with IPTG added at four concentrations: 0/0.1/0.2/0.4 mM (n=3). After protein induction, 1 mL of culture was taken, and the cells were harvested. The cells were then diluted in PBS to achieve an OD600 of 0.5. This sample was applied to a 96-well plate, and we measured fluorescence at 535 nm using excitation light at 485 nm using a microplate reader.
The results of the measurements are as shown in the table below.
| IPTG (mM) | 0 | 0.1 | 0.2 | 0.4 | PBS |
|---|---|---|---|---|---|
| RFU | 14761 | 30610 | 21519 | 24216 | 32 |
| RFU | 14854 | 28107 | 27298 | 23010 | 32 |
| RFU | 15319 | 31434 | 25030 | 22886 | 32 |
We performed Tukey's multiple comparison test to examine the fluorescence values.
Fluorescence intensity significantly increased upon IPTG induction compared to 0 mM (p < 0.05). The maximum expression was observed at 0.1 mM IPTG, while higher concentrations (0.2 mM and 0.4 mM) showed decreased fluorescence.
Fluorescence intensity measurements revealed the strongest fluorescence under conditions where 0.1 mM IPTG was added. Furthermore, GFP expression was confirmed even under conditions without IPTG addition (IPTG (-)), due to recombination occurring from Cre leak expression. However, the fluorescence intensity under these conditions was significantly lower compared to the IPTG-added condition. This indicates that the terminator designed between the loxP sites is functioning.
These results confirm that the loxP_NisinQ_sfGFP_Vector_Assembled plasmid constructed in Cycle 1 undergoes deletion of the sequence between loxP sites in vivo upon Cre recombinase action. Furthermore, it was shown that following recombination, IPTG induces expression of the downstream GFP.
By Cycle 3, we demonstrated that the constructed loxP plasmid undergoes recombination by Cre recombinase. Furthermore, by measuring fluorescence after deletion of the loxP-flanked region, we verified Cre-dependent protein expression and its functionality. In Cycle 4, we analyzed NisinQ—the segment positioned between the loxP sites that is deleted in a Cre-dependent manner. Notably, for NisinQ we adopted the NusA–NisinQ fusion part (BBa_K4437002) designed by iGEM Calgary 2022, considering molecular weight and solubility in *E. coli* (*2*).
For more detail about Human Practices, click here
We aim to confirm the expression of Nisin Q from the loxP plasmid using SDS-PAGE. We will also examine the difference in expression levels with and without IPTG induction.
After preculturing the loxP plasmid at 37°C, the cells were transferred to a main culture at 37°C. When the OD reached approximately 0.6, IPTG was added to a final concentration of 0.1 mM. Following incubation, 1 mL of the culture was collected. After centrifugation and resuspension, 500 μL of the cell extract was purified using affinity chromatography with a spin column, and then analyzed by SDS-PAGE.
As a result of the SDS-PAGE analysis of the loxP plasmid, a faint band was observed around 65 kDa, which corresponds to the expected molecular weight of the NusA–NisinQ fusion protein. Although very faint bands were also detected in the flow-through and wash fractions 1 and 2, the strongest band appeared in the elution fraction. Furthermore, since no other bands were observed in the elution fraction besides the one near 65 kDa, it is highly likely that this band represents the target fusion protein composed of NusA and NisinQ.
Although a band likely corresponding to the target protein was observed in the SDS-PAGE results, it was faint, and thus we could not conclusively confirm that it represents the NusA–NisinQ fusion protein. Unfortunately, due to time constraints, we were unable to perform additional experiments. However, increasing the amount of cell lysate applied to the column could enhance the detection of the target band, making it easier to evaluate both the expression level and the difference between IPTG-induced and non-induced conditions. To verify whether the observed band truly corresponds to the target fusion protein, further affinity purification using a larger amount of cells should be performed, followed by testing the isolated protein for antimicrobial activity in Wet lab experiments.
To prevent foodborne illnesses caused by bacteria on fresh produce, we evaluated the antimicrobial functionality of the natural peptide nisin, which cannot be handled in the laboratory, through simulation. Targeting Listeria monocytogenes, a pathogen capable of proliferating under refrigeration, we analyzed the effects of concentration, pH, and temperature by combining the Weibull model and the Hill function.
One of the concepts of this biosensor project is to prevent foodborne illness caused by bacteria attached to fresh produce. Therefore, we decided to evaluate the functionality of nisin, which serves to protect produce from bacterial contamination. Among Gram-positive bacteria that cause spoilage and food poisoning, we focused on Listeria monocytogenes. This pathogen can grow even under refrigerated conditions, posing a risk of reaching infectious levels in stored produce, and has been responsible for numerous outbreaks. However, since Listeria requires BSL-2 facilities and cannot be handled in our laboratory, we conducted a dry functional evaluation instead.
We aimed to quantitatively clarify how the bactericidal effect of the natural antimicrobial peptide nisin varies with three factors—concentration, pH, and temperature—by constructing a predictive model that integrates multiple mathematical models. Specifically, we used the Weibull model to describe the nonlinear decrease in bacterial count over time and applied the Hill function to reproduce the steep, switch-like response to changes in nisin concentration.
$$\boxed{\;\delta(C, \mathrm{pH}, T) \;=\exp (\; b_0 \;+\; b_{\mathrm{Hill}}\,H(C) \;+\; b_{\mathrm{pH2}}\,( \mathrm{pH}-\mathrm{pH}_{\mathrm{opt}} )^2 \;+\; b_T\,(T-T_{\mathrm{ref}}) \;)}$$
We defined parameters for concentration, pH, and temperature and estimated their optimal values. Parameter estimation was performed using the least squares method to minimize the residuals between observed and modeled values. The results showed that nisin is most effective under acidic conditions (pH ≈ 5.0) and at low temperatures, but requires relatively high concentrations to achieve sufficient antibacterial activity.
Interpreting these results in the context of our project, nisin works more effectively on produce that tends to be slightly acidic. Although Listeria monocytogenes can proliferate even in refrigerated environments, nisin can help prevent such foodborne risks during cold storage. Furthermore, the study demonstrated that effective sterilization requires high concentrations of nisin (10²–10³ µg/mL) over extended periods. Through the process of building this multivariate model, we realized that the model’s accuracy strongly depends on the quality and diversity of input data, reaffirming the importance of robust data collection in biosensor development.
Interpreting these results in the context of our project, nisin works more effectively on produce that tends to be slightly acidic. Although Listeria monocytogenes can proliferate even in refrigerated environments, nisin can help prevent such foodborne risks during cold storage. Furthermore, the study demonstrated that effective sterilization requires high concentrations of nisin (10²–10³ µg/mL) over extended periods. Through the process of building this multivariate model, we realized that the model’s accuracy strongly depends on the quality and diversity of input data, reaffirming the importance of robust data collection in biosensor development.
Through Cycle 2-4, we successfully verified the functionality of the constructed loxP plasmids. In previous in vivo analyses, a plasmid constitutively expressing Cre recombinase was used. However, to achieve the goals of Protection, Sensing, and Visualization in this project, it is necessary to employ a Cre expression plasmid that responds to ethylene. Therefore, we used the EtnR1R2_Petn_Cre_GoldenGate plasmid constructed in Cre Cycle 3 to co-transform BL21 cells, thereby generating the Sense strain for this system.
The EtnR1R2_Petn_Cre_GoldenGate plasmid encodes EtnR1 and EtnR2, transcription factors that are activated by ethylene metabolites, as well as their corresponding binding promoter, Petn. Theoretically, this plasmid enables Cre expression induced by epoxyethane, a metabolite of ethylene. For this purpose, the Cre plasmid and the loxP plasmid were co-transformed into E. coli BL21(DE3).
Cotransformation was performed using the EtnR1R2_Cre plasmid and the loxP plasmid. To prevent plasmid incompatibility, the two plasmids were designed with different origins of replication. The former carries an ampicillin resistance gene, while the latter carries a chloramphenicol resistance gene. Therefore, by culturing on medium containing both ampicillin and chloramphenicol, colonies that successfully underwent cotransformation were reliably screened and selected. To convert ethylene into epoxyethane, which can activate EtnR1R2, a monooxygenase-expressing strain was employed. Accordingly, a coculture simulation was conducted using the constructed strain together with the monooxygenase-expressing strain.
Transformation was successful, and colonies were obtained. These colonies were preserved as glycerol stocks, designated as the Sense strain. The results of the coculture simulation conducted with the Sense strain and the monooxygenase-expressing strain are presented below.
For more detail about Dry project 1, click here
At this stage, the functional analysis of EtnR1R2_Cre has not yet been completed; therefore, the experiment will not be conducted. However, it will be possible to perform an in vivo Cre recombination assay following the Cycle 3 protocol, as well as to use this strain for coculture simulations with monooxygenase-expressing E. coli that oxidizes ethylene.
We have not yet demonstrated the expression and function of EtnR1R2, and resolving this is a priority. Once completed, we will perform validations analogous to Cycle 3 to determine whether environment-responsive Cre, induced via the EtnR1R2–Petn module, mediates recombination at the loxP sites in vivo.
With the design goal of developing a module that enables the Tom protein (monooxygenase) to function as an ethylene reactor—converting ethylene into epoxyethane detectable by the sensor strain—we aim to establish an integrated DBTL cycle combining wet experiments and dry modeling.
Specifically, we will first determine the DNA sequence of Tom derived from Burkholderia cepacia G4 and redesign it for expression in E. coli. Based on this optimized sequence, we will conduct plasmid construction as part of the design phase. Next, we will attempt wet-lab experiments to verify Tom expression; however, if expression is difficult to confirm experimentally, we will employ computational modeling to predict Tom expression levels and evaluate expected enzyme activity through in silico simulations.
Subsequently, we will integrate the Tom-expressing strain into a co-culture model with the epoxyethane sensor strain, allowing us to simulate both ethylene oxide (Eto) accumulation and sensor response following ethylene addition. Through this iterative process, we will validate and optimize the Tom expression module, ultimately establishing a robust engineering cycle that integrates the module into the ethylene-responsive biosensor system.
In this project, the V106F and A113F mutants of toluene/o-xylene monooxygenase (Tom) were selected as target enzymes functioning as ethylene monooxygenases. However, since sequence information for both mutants was unavailable, the related gene cluster was investigated, and the mutation sites were identified based on literature data. Mutations were then introduced into the wild-type sequence accordingly. Subsequently, codon optimization was performed based on the codon usage bias of the host organism, and the desired mutant sequences were constructed.
In this project, the V106F and A113F mutants of toluene/o-xylene monooxygenase derived from Burkholderia cepacia G4 were selected to function effectively as ethylene monooxygenases. However, since the sequence information of these mutant enzymes was not available in existing databases, the gene cluster sequence of Burkholderia cepacia G4 Tom (TomA0-TomA5) was first investigated using GenBank (*1). Subsequently, to determine which subunit corresponded to the V106F and A113F mutations, a comparative analysis was conducted, revealing that only TomA3 contained valine at position 106 and alanine at position 113 (Figure 1).
Figure 1: Comparison of the amino acid sequences of TomA1, TomA3, and TomA4 at residues 106 and 113
Therefore, nucleotide sequences of the enzyme genes obtained from GenBank were modified to introduce either the V106F or A113F mutation. These sequences were then optimized using VectorBuilder (2) according to the codon usage bias of Escherichia coli (strain K-12, substr. MG1655), while ensuring that no restriction enzyme sites were included.
Through sequence optimization using VectorBuilder, the Codon Adaptation Index (CAI) and GC content were obtained as shown in Table 1. This optimized sequence was used for the prediction of expression levels by machine learning.
Table 1. Evaluation of CAI and GC content after optimization.
| Gene | GC Content (%) | CAI |
|---|---|---|
| TomA0 | 54.22 | 0.98 |
| TomA1 | 59.04 | 0.94 |
| TomA2 | 48.52 | 0.97 |
| TomA3 of V106F | 50.32 | 0.95 |
| TomA3 of A113F | 50.26 | 0.95 |
| TomA4 | 59.10 | 0.96 |
| TomA5 | 55.68 | 0.94 |
Figure1. the whole sequence of Tom A113F Mutant
Figure2. the whole sequence of Tom V106F Mutant
In the sequence determination process of this study, the gene cluster of Tom derived from strain G4 was found to contain six coding regions, designated TomA0 to TomA5. According to previous studies, the functional enzyme consists of three subunits—TomA1, TomA3, and TomA4—which assemble into an (αβγ)₂ structure to exhibit enzymatic activity. Based on this information, it was assumed that the mutations occurred in TomA3. In fact, similar mutations have been reported in TomA3 in earlier studies, supporting the validity of this assumption (※3, 4).
EtnR1/EtnR2 are transcription factors that do not respond to ethylene itself but react to its monoxide, epoxyethane. Therefore, an enzyme is required to oxidize ethylene released from fruits and vegetables into epoxyethane. In this study, we created the plasmid TOM_A113F_GoldenGate by codon-optimizing the ethylene monooxygenase Tom and introducing the A113F mutation. TOM_TomA0124_Vector and TOM_A113F_Tom34_BsaI were ligated via Golden Gate Assembly. Electrophoresis analysis confirmed a shifted-up band compared to the vector, demonstrating successful construction of the TOM_A113F_GoldenGate plasmid.
EtnR1/EtnR2 are transcription factors that do not respond to ethylene itself but react to its monoxide, epoxyethane. Therefore, an enzyme is required to oxidize ethylene released from fruits and vegetables into epoxyethane. In this study, we created the plasmid TOM_A113F_GoldenGate by codon-optimizing the ethylene monooxygenase Tom and introducing the A113F mutation. TOM_TomA0124_Vector and TOM_A113F_Tom34_BsaI were ligated via Golden Gate Assembly. Electrophoresis analysis confirmed a shifted-up band compared to the vector, demonstrating successful construction of the TOM_A113F_GoldenGate plasmid.
Recombination was performed using the TOM_TomA0124_Vector plasmid and the TOM_A113F_Tom34_BsaI plasmid via Golden Gate Assembly. To confirm the success of the recombination, agarose gel electrophoresis was performed, and it was determined that the band would shift upward compared to the vector (TOM_TomA0124_Vector plasmid).
*TOM_TomA0124_Vector plasmid = 6015 bp, TOM_A113F_GoldenGate plasmid = 7194 bp
The electrophoresis results showed that the TOM_A113F_GoldenGate band was positioned above the TOM_TomA0124_Vector plasmid band, confirming the expected shift-up.
Based on the electrophoresis results, we determined that the TOM_A113F_GoldenGate plasmid was successfully constructed.
Using the plasmid constructed in Cycle 1, we attempted to confirm the expression of TomA0-A5. However, since the protein was not tagged or otherwise labeled, we were unable to detect a clear band. Therefore, we decided to proceed to expression simulation in Dry.
The Tom complex consists of six subunits (TomA0-A5). Therefore, with the cooperation of Prof. Yokogawa, SDS-PAGE was performed to confirm protein expression from the plasmid created in Cycle 1-2.
E. coli cells induced with IPTG (A: 0.2 mM or B: 0.4 mM) were disrupted, and the soluble and insoluble fractions were applied separately. Since the molecular weights of TomA0-A5 are widely distributed, gels prepared as follows were used.
We've identified several bands, but we are not entirely sure which Tom they correspond to.There are too many bands to see clearly. We should have tagged each protein.
TomA0 = 8.3 kDa
TomA1 = 37.5 kDa
TomA2 = 10 kDa
TomA3 = 61.0 kDa
TomA4 = 13.1 kDa
TomA5 = 39.2 kDa
We attempted to confirm TOM expression using the constructed plasmid, but were unable to detect expression of any of the six TOMs. Based on this result and the professor's advice that it is difficult to perfectly confirm the expression of the six untagged proteins, we decided to proceed with in vitro expression simulation.
The biosensor proteins EtnR1 and EtnR2 used in this project respond specifically to ethylene oxide and are therefore unable to directly detect ethylene. Consequently, to achieve ethylene sensing, it is necessary to introduce and express an enzyme that converts ethylene into ethylene oxide in a separate Escherichia coli strain from the one expressing EtnR1 and EtnR2. The enzyme employed in this study is a mutant of toluene/o-xylene monooxygenase derived from Burkholderia cepacia G4. Since few previous studies have investigated this mutant, and its expression level in E. coli remains unknown, a Python-based prediction program was developed to estimate its expression level.
In this study cycle, it was confirmed that the V106F and A113F mutants of toluene/o-xylene monooxygenase (Tom) function effectively as ethylene monooxygenases (※3-1). Therefore, the nucleotide sequences of both mutants were constructed using information from literature and databases, and a Python-based program was developed to predict their expression levels. As training data, protein expression levels and the corresponding DNA sequences of Escherichia coli were obtained from PaxDb (※3-2) and UniProt (※3-3).
As mentioned above, in developing the prediction program, the approach was designed based on information derived from nucleotide sequences that encode the target proteins. However, since the information obtainable from nucleotide sequences (e.g., the secondary structure of mRNA, free energy and diversity in the thermodynamic ensemble, and mountain plots, etc.) is highly diverse, only a selected subset of these features was used for the prediction process.
To obtain nucleotide sequence-derived information from the input DNA sequences, the ViennaRNA package was implemented in the Python code to calculate the “minimum free energy (MFE)” and “minimum free energy structure (SS)” of the DNA sequences obtained from the database(※3-2, 3-3). In addition, the Longformer model was employed to extract features from the RNA sequences. These extracted data were then combined with the corresponding “expression levels” into a single DataFrame, which was prepared as the training dataset.
The dataset was divided into features (X) and the target variable (y: expression level), and then split into training and test sets. A random forest regression model was trained, and the predictive performance on the test data was evaluated using the mean squared error (MSE). Subsequently, RNA sequences were generated from unknown DNA sequences, and the expression levels were predicted through the same procedure as for sequences with known expression levels. As a result, when predictions were performed in two stages using 40 and 95 training data points, respectively, the MSE was smaller with 95 data points, and the gap between the “observed” and “predicted” values was reduced.
Initially, when obtaining protein expression levels of E. coli and their corresponding DNA sequences from the database, the data were entered in descending order of expression levels. As a result, when the training dataset contained 40 samples, a clear tendency of Predicted > Observed (Actual) was observed in the scatter plot. Therefore, to improve the prediction accuracy and enable comparison, 57 additional pairs of DNA sequences and expression data were added, and the two sequences with the highest expression levels were removed as outliers before re-performing the prediction. As a result, a decrease in MSE was observed, and the scatter plot showed that the relationship between expression levels became closer to Predicted ≒ Observed (Actual). These findings suggest that, in addition to increasing the number of data points and removing outliers, randomly including DNA sequences with significantly different expression levels is also important for improving prediction accuracy. Furthermore, since the enzyme showed a consistent level of expression in all predictions, the next step was to proceed to the functional evaluation of this enzyme.
Building upon the results of Cycle 3, we formulated the rate equation describing the change in concentration (formation rate) of ethylene oxide, the product of the enzymatic reaction catalyzed by TOM. This equation was implemented in Python to simulate the concentration dynamics, and the functionality of the enzyme was evaluated primarily based on the behavior of these concentration changes.
From the results of Cycle 3, we confirmed the expression of TOM in E. coli. Based on this outcome, we conducted a simulation to evaluate the functionality of the expressed TOM enzyme and to determine whether it could serve as a practically useful biocatalyst in our project.
Further literature research was conducted; however, we were unable to find a detailed enzymatic reaction equation describing the catalytic process inside TOM. Therefore, we assumed that the enzymatic reaction occurring within TOM proceeds through a mechanism similar to that of alkene monooxygenases (11), and constructed a corresponding reaction model.
Based on this assumption, we derived rate equations describing the concentration changes of the product ethylene oxide (C₂H₄O) and the substrate ethylene (C₂H₄).
To verify the validity of this assumption, we consulted Professor Atsuhiro Shimada at Gifu University as part of our Human Practices activities.
A Python simulation was developed based on the constructed rate equations and fitted to experimental data reported in the literature (※3-1). The simulation was executed, and the results can be viewed on the Dry Lab’s Model 1 page via the link below.
Example of simulation results:
・Expressed enzyme: A113F
・Expressed enzyme: V106F
The obtained simulation results demonstrated that both TOM variants, A113F and V106F, exhibited time-dependent concentration changes consistent with a typical first-order enzymatic reaction. However, the reaction time for A113F was approximately 250 minutes, whereas V106F required around 8,000 minutes—roughly 32 times longer.
These findings suggest that, within the experimental framework of our project, the A113F variant—with its significantly shorter reaction time—is the more suitable enzyme for practical application.
In this project, we developed and analyzed a mathematical model describing the behavior of a co-culture system composed of Escherichia coli engineered to express monooxygenase, which converts ethylene into ethylene oxide, and sensor E. coli responsive to ethylene oxide. First, using single-strain cultivation data, we estimated fundamental parameters such as the maximum specific growth rate and the substrate half-saturation constant. Next, enzyme expression levels and catalytic turnover numbers were estimated, and the epoxidation reaction was described using the Michaelis-Menten framework. Furthermore, growth inhibition by the product, ethylene oxide, was incorporated into the model. Substrate consumption, product accumulation, and sensor response under co-culture conditions were then represented as a system of ordinary differential equations (ODEs). Through this model, we simulated the dynamics under different initial biomass and substrate concentrations and identified conditions that maximize the sensor response.
This study aimed to model the dynamics of a co-culture system consisting of a monooxygenase-expressing strain that oxidizes ethylene to ethylene oxide and a sensor strain that responds to the produced ethylene oxide. First, single-strain cultivation experiments were designed to obtain data for estimating key Monod parameters, including the maximum specific growth rate (μₘₐₓ), the substrate half-saturation constant (Kₛ), and the yield coefficient (Yₓ/ₛ).
Based on these values, we constructed an integrated system of ordinary differential equations (ODEs) describing the growth of both strains, substrate consumption, the ethylene oxidation reaction (modeled by the Michaelis–Menten equation), and the growth-inhibitory effects of the product. The resulting unified model was designed to predict the time-course fluorescence output of the sensor strain (SensorOutput). Furthermore, the model can be applied to exploring optimal conditions for maximizing sensor response and to parameter sensitivity analyses.
The model was implemented in Python, employing SciPy’s solve_ivp function for numerical integration and scipy.optimize.curve_fit for parameter estimation. First, Monod parameters were fitted using single-strain cultivation data as input, and the resulting parameters were incorporated into the co-culture model.
For the enzymatic reaction, values of Vₘₐₓ and Kₘ were derived from literature data, which were then used to calculate kₐₜ and the corresponding enzyme concentration. These parameters were integrated into the model as part of the reaction rate expression. To account for inhibitory effects and product accumulation, a function was introduced in which the growth rate μ is suppressed as a function of the ethylene oxide concentration (P).
The entire model was modularly structured, allowing independent validation of each process—biomass growth, substrate consumption, product formation, and sensor output—thus enabling flexible testing and refinement of individual system components.
Fig.1 Ordinary Differential Equation (ODE) system representing the co-culture dynamics after ethylene addition
Using the constructed model, simulations were performed under both single-culture and co-culture conditions, and the resulting biomass dynamics, substrate consumption, product concentration, and sensor responses were analyzed.
Furthermore, optimization of the initial conditions (S_0, X_{0,1}, X_{0,2}) was conducted, with the maximum sensor output y′_max defined as the objective function. An initial grid search was employed to capture the overall trend, followed by Bayesian optimization using Optuna to improve search precision.
As a result, it was confirmed that variations in the initial substrate concentration and cell ratio significantly affected the sensor response, enabling the quantitative identification of conditions that maximize response intensity.
Fig.2 Comparison of Sensor Outputs
Simulation results revealed that enzyme expression level, initial cell ratio, and substrate concentration strongly influence the sensor response. Furthermore, the difference in growth rates between the TOM-expressing strain and the sensor strain was suggested to contribute to product accumulation and response delay.
However, several simplifications remain in the current model. In particular, the model does not account for induction delay of enzyme expression, metabolic burden of protein synthesis, gas–liquid equilibrium and diffusion losses of ethylene and ethylene oxide, or intracellular localization effects. In addition, the sensor response was approximated using a Hill equation, which does not capture the nonlinear dynamics or stochastic noise inherent in transcription factor behavior.
Future work will involve extending the model to incorporate these factors and validating parameters through experimental comparison. This refinement is expected to enable quantitative design of biosensors suitable for real-world applications such as fruit ripening detection and other environmental ethylene-sensing systems.
Back to Top