Engineering Success: bile acid sensor utilizing the VFA0359 repressor

Background

Our engineered probiotic platform is designed to monitor intestinal microbiome metabolite concentrations in real time and respond accordingly. However, metabolite levels within the intestinal lumen can fluctuate throughout the day due to the host's dietary intake and intestinal movements. For instance, low butyrate levels may be expected in places where food residues are absent, as the intestinal microbiome lacks a carbon source for metabolite production. Therefore, our probiotic must differentiate between the presence and absence of food residues when responding to metabolite levels. Bile acids, released by the gallbladder during digestion, can serve as an indicator of food presence. To address this, we plan to incorporate a biosensor capable of detecting and responding to changes in bile acid concentrations into our probiotic platform. This will enable our probiotic to regulate the gut metabolites profile based on real-time environmental signals effectively.

Design (Cycle #1)

After research, we selected the bile acid sensor designed by Mao Taketani et al. (2020) for integration into our probiotic. This biosensor has low basal activity, high dynamic range, and no observable cross-talk with other common sensor systems, which makes it an ideal choice for our platform (Mao Taketani et al., 2020). This bile acid sensor was constructed in Bacteroides thetaiotaomicron by placing the gene of the TetR-family repressor VFA0359 under the control of a constitutive promotor and placing the operator sequence between the −33 and −7 region of a constitutive promoter that regulates the expression of a reporter gene (Figure 1). The biosensor responds to multiple kinds of bile acids but showed highest response to deoxycholic acid (DCA), a representative compound in bile acids (Mao Taketani et al., 2020). Thus, we chose DCA as our inducer to test the induction of our bile acid sensor based on the VFA0359 repressor.

Figure 1. The genetic circuit of the bile acid sensor designed by Mao Taketani et al. Created by biorender.com.

Before obtaining the sequence of this bile acid sensor and testing it in our probiotic, E. coli Nissle 1917, we first ran an in silico stimulation of the transcription Initiation rates of this sensor in Escherichia coli str. K-12 substr. MG1655 using the Promoter Calculator function of De Novo DNA (LaFleur, T. L., 2022). The results (Figure 2) showed that the transcriptional activity of PcfxA is too low (blue box and arrow) and that there are numerous unwanted transcription start sites. This could be because this bile acid promotor is designed for B. thetaiotaomicron instead of E. coli.

Figure 2. De Novo DNA Promotor Calculator result of the bile acid sensor designed by Mao Taketani et al. in comparison with the sequence of the sensor. Sequence obtained from plasmid pMT447 (Addgene Plasmid #139387).

Due to the low desired transcriptional activity and numerous unwanted transcription start sites of this bile acid sensor, we decided to engineer this bile acid sensor to fit the transcription and translation environment of E. coli. VFA0359 used in this bile acid sensor belongs to the TetR-family repressors, the same family as the repressor EilR in the Jungle Express repressor system. The Jungle Express repressor system is a tight transcriptional control system built for E. coli by Thomas L. Ruegg et al. (2018). We used this repressor system in our 2022 project (BBa_K4204016). Considering the structural similarity and compatibility shared between proteins within the same family, we decided to use the framework (everything except the repressor and operator) of the jungle express system as a substitute for the framework of the DCA sensor system. We adopted SnowPrint (d’Oelsnitz, S. et al., 2024) to predict the palindromic operator sequence of VFA0359 and inserted two operator sequences in the phage early promoter, one overlapping with -10/-35 region and another one downstream -10 hexamer. The resulting new design and De Novo DNA Promoter Calculator (LaFleur, T. L., 2022) results of the new design are shown in Figures 3 and 4. The transcription rate peaks in regions with blue boxes and arrows show that this design has a desired transcriptional activity at promoters driving the expression of the VFA0359 regulator and the target gene (GFP_mut2).

Figure 3. New design of the bile acid sensor utilizing the Jungle Express framework. Created by biorender.com.

Figure 4. The new design of the bile acid sensor utilizing the Jungle Express framework and its De Novo DNA Promotor Calculator result. The blue boxes and arrows highlight the desired regions of transcriptional activity.

See more details on design page.

Build (Cycle #1)

The bile acid sensor design is then integrated into the plasmid backbone pUC57, the codons of the two proteins were optimized for expression in E. coli using algorism provided by Genscript, and the plasmid (pUC57_Kan_pDCA_VFA0359) was then obtained from Genscript (Plasmid map see Figure 5).

Figure 5. Plasmid map of pUC57_Kan_pDCA-VFA0359.

Test (Cycle #1)

After Obtaining the plasmid pUC57_Kan_pDCA_VFA0359 from Genscript, we transformed it into E. coli trelief 5α and performed a kinetic assay to characterize our new bile acid sensor design (detailed protocol of kinetics assay are in protocol page). Overnight cultures of pUC57_Kan_pDCA_VFA0359 transformed trelief 5α cells in LB media at 37°C were 1:250 (v/v) inoculated into fresh M9 media (M9 media with 0.4% (w/v) glucose, 0.2% (w/v) casamino acid, and 1 mM thiamine, the same below), cultured for 3 hours and diluted to OD=0.01 using M9 media. The resulting culture was then separated into aliquots, and different concentrations of DCA were added to the aliquots. Then, the aliquots were plated onto a well black/clear bottom 96-well plate with three repeats each and 150 µL culture in each plate. Positive/negative control E. coli cultures with different plasmids were also processed using the same above-mentioned method. The positive control used is a J23100-sfGFP on a pSB1C3 backbone (BBa_K515105). The negative control used is a tes4 gene on a pET-28a(+) backbone (BBa_K3838613). The detailed plate setup is shown below in Table 1, and the kinetics assay was performed as described in the protocol.

Table 1. Plate layout of kinetics assay.

The results of the kinetics assay were used to calculate the fluorescence/ABS600 (GFP/OD) curve. The GFP/OD curve of the groups shows that there are no significant differences in GFP expression levels between the groups with different concentrations of DCA (Figure 6). This suggests that our biosensor design did not function as we hypothesized.

Figure 6 .The kinetics assay result of the new bile acid sensor, plasmid pUC57_Kan_pDCA_VFA0359. Unit of time is seconds.

We also compared the fluorescence value of our bile aid sensor pUC57_Kan_pDCA_VFA0359 with the positive control, which is a constitutively expressed GFP on a pSB1C3 backbone. The results show that our biosensor has a very low level of GFP expression compared to that of a control.

Figure 7. The fluorescence value graphed against time of the new bile acid sensor on the plasmid pUC57_Kan_pDCA_VFA0359 and of the positive control. Unit of time is seconds, and "2002p" is the plasmid that has a constitutively expressed GFP on a pSB1C3 backbone.

Learn (Cycle #1)

The results show that our sensor is not able to report the difference between different concentrations of DCA and that the basal expression rate of our sensor system is fairly low. Given that there is proof that the repressor VFA0359 can bind to the operator and that DCA can release the repressor from the operator (Mao Taketani et al., 2020), Several factors could contribute to the observed result. One possibility is that the VFA0359 repressor may be overexpressed and/or the repressor has an exceptionally strong binding affinity for the operator. This could hinder the ability of DCA to effectively release the repressor from the operator, preventing the activation of the output promoter. Alternatively, DCA might be unable to reach the repressor to exert its result. Also, the output promoter, after the addition of the two operators, could malfunction, failing to initiate GFP expression even when the repressor is released. Further investigation is necessary to pinpoint the cause of the observed results.

Design (Cycle #2)

As pointed out in the "learn" part of the first cycle, one possible reason for the malfunctioning of our bile acid sensor could be because of the output promoter. Thus, in this cycle, we aim to test whether the output promotor, O1O2+PjexD, or PVFA0359, functions as normal. We decided to mutate the repressor VFA0359 to stop its expression by changing the two codons after the start codon to the stop codon TAA. We hypothesize that after this silencing mutation to the VFA0359 repressor, the GFP expression level of E. coli with the plasmid pUC57_Kan_pDCA_VFA0359 would significantly increase, which suggests that the output promotor functions as desired and that VFA0359 does exert a repression effect on the transcription initiation activity of the output promotor. The mutated plasmid is named pUC57_Kan_pDCA_VFA0359_2TAA.

Figure 8. The difference between pUC57_Kan_pDCA_VFA0359 (Top) and pUC57_Kan_pDCA_VFA0359_2TAA (bottom).

See more details on design page.

Build (Cycle #2)

PCR and Golden Gate assembly were performed to achieve the change in sequence. Agarose gel electrophoresis was performed and the bands indicate the correctness of the components' length (Figure 9).

Figure 9. The AGE result of the PCR product for the construction of plasmid pUC57_Kan_pDCA_VFA0359_2TAA The bands indicate the correctness of each component's length. pUC57_Kan_pDCA_VFA0359_2TAA is abbreviated to pDCA-2TAA.

Test (Cycle #2)

The plasmid pUC57_Kan_pDCA_VFA0359_2TAA was transformed into E. coli trelief 5α. Figure 10 shows the colonies that are successfully transformed with pUC57_Kan_pDCA_VFA0359_2TAA (with a yellow circle) exhibit fluorescence under blue light. The single colonies were inoculated in LB media and cultured overnight at 37 °C and 220 rpm. The overnight cultures were then centrifuged at 3000 g to obtain cell pellets to observe fluorescence. The results show that pUC57_Kan_pDCA_VFA0359_2TAA has a significantly higher basal fluorescence than pUC57_Kan_pDCA_VFA0359 and is close to that of pJump47-2A (Addgene Plasmid #126991). However, the picture of the result of this experiment is missing at the time when writing this wiki page. Our team will experiment again and update our results.

Figure 10. LB Agar plates with colonies of pUC57_Kan_pDCA_VFA0359_2TAA (right) and pJump47-2A (left) under the blue light of a blue LED transilluminator. The circles indicate colonies successfully transformed with pUC57_Kan_pDCA_VFA0359_2TAA (verified by Sanger sequencing).

Learn (Cycle #2)

Our results suggest that the output promoter, PVFA0359, with the two operators inserted, could initiate GFP transcription as we desired. This also provides proof that VFA0359 repressed the expression of GFP in our previous bile acid sensor design. Thus, the possible explanation regarding our bile acid sensor pUC57_Kan_pDCA_VFA0359 not functioning is that the VFA0359 repressor may be overexpressed and/or the repressor has an exceptionally strong binding affinity for the operator. To solve this problem, we decided to weaken the expression of VFA0359 and weaken its binding with the output promotor in the next engineering cycle.

Design (Cycle #3)

To weaken the expression of VFA0359 and weaken its binding with the output promotor, we planned to engineer the RBS that regulates the translation of VFA0359 for weaker translation initiation and delete the first of the two operators on the output promoter. The original GSG RBS used to express the repressor EliR in the Jungle express system was mutated in silico by the RBS calculator on De Novo DNA (Salis, H. M.,2009). A mutant with a translation initial rate of 1000 au was selected from multiple mutants and named GSG RBSmut1000. We deleted the operator in between the -35 and -10 motif and replaced the -10 motif with a stronger one. The resulting plasmid after the two changes is named pUC57_Kan_pDCA_VFA0359_mut1000_operator (abbreviated to pDCA_mut1000_operator)(Figure 11). The two changes are shown in Figure 12.

Figure 11. The genetic circuit of the new bile acid sensor design pDCA_mut1000_operator. Created by biorender.com.

Figure 12. Comparison of the changes made in the design of this cycle. "Before" refers to the plasmid pUC57_Kan_pDCA_VFA0359 and "After" refers to the plasmid pUC57_Kan_pDCA_VFA0359_mut1000_operator.

See more details on design page.

Build (Cycle #3)

We designed primers to perform PCR on the original plasmid pUC57_Kan_pDCA_VFA0359 and obtained fragments with the desired changes in sequence. Then, we utilized Golden Gate Assembly to assemble the new plasmid pDCA_mut1000_operator from the fragments.

Test (Cycle #3)

Upon successful Golden Gate Assembly of the plasmid pDCA_mut1000_operator, we transformed it into E. coli trelief 5α. Single colonies of pDCA_mut1000_operator and pUC57_Kan_pDCA_VFA0359 were picked from LB agar plates and inoculated in LB media with 1000 µM, 200 µM, or 0 µM DCA and cultured overnight at 37 °C and 220 rpm, each with three repeats. The overnight cultures were plated onto wells of a well black/clear bottom 96-well plate and placed into the microplate reader to obtain Abs600 and fluorescence value (settings of the microplate reader are the same as the kinetics protocol). The average net fluorescence/ABS600 (FL/ABS) values were calculated and plotted in Figure 13. The results showed that the overnight culture of pDCA_mut1000_operator exhibited a GFP expression level even lower than before (compared to pUC57_Kan_pDCA_VFA0359). This suggests that there must be some flaws in our design.

Figure 13. The average net FL/ABS value of overnight cultures of pDCA_mut1000_operator and pUC57_Kan_pDCA_VFA0359 in E. coli Trelief.

Learn (Cycle #3)

To better identify the problem shown in the results section of the plasmid design pDCA_mut1000_operator, we used De Novo DNA promoter calculator (LaFleur, T. L., 2022) to calculate transcription initiation rates of the sensor sequence of pDCA_mut1000_operator. The results show that there is a missing forward transcription peak at the position of the PjexD promoter, and there is an unwanted peak above the RiboJ-RBS-GFP region.

Figure 14. The bile acid sensor on pDCA_mut1000_operator and its De Novo DNA Promotor Calculator result. The red box indicates a missing peak, and the blue box indicates an unwanted peak.

We looked into the details in sequence view to find the reason for the missing and unwanted peak. It turns out that the missing peak is because we accidentally deleted one bp, an adenine, from the -35 motif of the PVFA0359 promotor (see Figure 14) when removing the operator and did not realize it. The unwanted peak is due to the codon optimization done in the "build" section of cycle 1 of the GFP gene, which uses 5'-TCA-3' instead of the original 5'-AGT-3' for the second amino acid serine and creates a strong -35 motif at the start of GFP (see Figure 15). This strong -35 motif, together with the -10 motif on the preceding RBS, creates the unwanted peak that we see in Figure 15.

Figure 15. The unwanted promoter and its -35 and -10 motifs were created accidentally by codon optimization.

Design & Build (Cycle #4)

To correct the two mistakes identified in the learn section above, we designed two sequential Golden Gate Assemblies, the first one fixing the missing "A" on the -35 motif of the output promotor (plasmid called pDCA_Fix_1), and the second one changing the codon that codes for the second amino acid, serine, back to 5'-AGT-3' (plasmid called pDCA_Fix_2).

Figure 16. The AGE result of the PCR products for the Golden Gate Assemblies of the plasmids pDCA_Fix_1 (A) and pDCA_Fix_2 (B). The bands indicate the correctness of each component's length.

See more details on design page.

Test (Cycle #4)

Upon successful Golden Gate Assembly of the plasmid pDCA_Fix_2, we transformed it into E. coli Nissle 1917. Single colonies of pDCA_Fix_2 were picked from LB agar plates and inoculated in LB media overnight. GFP fluorescence kinetic assay was done using the pDCA_Fix_2 EcN overnight culture following the steps in the protocol. The detailed plate setup is shown below in Table 2.

Table 2. Plate layout of kinetics assay for pDCA_Fix_2.

The results of the fluorescence kinetic assay (Figure 17) show that higher DCA concentration in media results in stronger GFP expression levels. This suggests that our bile acid sensor utilizing the VFA0359 repressor succeeded in detecting and reporting changes in bile acid concentration quantitatively. The dynamic range of this sensor system (pDCA_Fix_2) at 500 µM was about 2.2-fold at 1000μM DCA, which is relatively low.

Figure 17. The fluorescence kinetics assay result of the plasmid pDCA_Fix_2.

Learn (Cycle #4)

Our sensor design finally succeeded at detecting changes in bile acid (DCA) concentrations. However, there are many future experiments and possible improvements. Repeats of the fluorescence kinetics assay could be done with another positive control, pUC57_Kan_pDCA_VFA0359_2TAA, to predict the potential upper limit of the dynamic range of our sensor design. Then, optimizations could be done regarding the expression of the VFA0359 repressor, the sequence and position of the VFA0359 operator, and the strength of the output promotor, PVFA0359, to enlarge the current dynamic range (or control it to reach a desired value) of our bile acid sensor.

Engineering success: SCFA biosensor

Introduction

Short-chain fatty acids (SCFA), such as propionate and butyrate, produced by the gut microbiota, are crucial metabolites. Microbiome dysbiosis, leading to changed SCFA profiles, is linked to some diseases, such as IBD, which is marked by a drop in butyrate levels and heightened intestinal inflammation. Our project aimed to detect the insufficiency in levels of SCFA in the human gut and compensate for it by producing SCFA. We plan to achieve this by designing a genetic circuit that links SCFA-sensing biosensors to a "NOT" gate. This circuit would monitor SCFA levels and express Tes4, an enzyme responsible for SCFA production, inversely proportional to SCFA levels, to compensate for the insufficiency in SCFA. The work of previous iGEM teams provided us choices for SCFA sensors, one of which is PpchA promoter that comes from the registry BBa_K4442001 (iGEM23_NMU-China).  Here, we engineer this SCFA sensor with the cI repressor system(NOT gate) together to create a functional genetic circuit.

Cycle #1: PpchA-cI biosensor validation

Design (Cycle #1)

Ppcha/Lrp is a butyrate biosensor system originating from Escherichia coli strain O157:H7 (Kineret et al., 2023). The pPcha/Lrp system has already been tested by previous iGEM teams, confirming it could be activated with the presence of butyrate (iGEM23_NMU-China).

We plan to examine the efficiency of the Ppcha sensor system by engineering E. coli and expressing it under controlled conditions. To achieve this goal, we designed plasmid pPcha_cI. In our design, leucine-responsive regulatory protein (Lrp) is constitutively expressed using the J23101 promoter, while plam controls expression of desired output genes. When butyrate is not present, plam express genes successfully. When Lrp binds to the pPcha promoter with presence of butyrate, it activates the expression of cI regulator, which then inhibits promoter pLam, and stops expression of output genes.

Figure 18. The genetic circuit of the pPcha_cI butyrate sensor. Created by biorender.com.

Figure 19. Plasmid map of pUC57_Kan_pPcha_cI.

See more details on design page.

Build (Cycle #1)

We integrated this design into the plasmid backbone pUC57-Kan, and ordered the plasmid from a commercial company. However, the plasmid synthesized by the company contained a point mutation on the ribosome binding site (RBS), which might effect the expression of Lrp.

Figure 20. Point mutation on the RBS for Lrp expression.

Test (Cycle #1)

We transformed pUC57_Kan_pPcha_cI into E. coli Trelief 5α. Single colonies of pUC57_Kan_pPcha_cI were picked from LB agar plates and inoculated in LB media overnight. GFP fluorescence kinetic assay was done using the pUC57_Kan_pPcha_cI Trelief overnight culture following the steps in the protocol. The detailed plate setup is shown below in Table 3.

Table 3. Plate layout of kinetics assay for pUC57_Kan_pPcha_cI.

The results of the kinetics assay were used to calculate the fluorescence/ABS600 curve. The normalized curve of the groups showed slight differences and a correct trend as the gradient of inducer was added, though the dynamic range was suboptimal (Figure 21).

Figure 21. Kinetics of PpchA before fixing with multiple butyrate concentrations over 18 hours. Fluorescence / ABS was used to represent GFP expression; higher fluorescence represented lower PpchA activity. The butyrate concentration ranges from 0 to 70mM.

Learn (Cycle #1)

From the results we found that there isn't significant induction of the biosensor by addition of SCFA butyrate. Given that the effect of the system has been proven by previous iGEM teams, one possible source of error could be the point mutation on RBS of Lrp, affecting its expression so the dynamic range is low. To address this, we utilized the promoter calculator function by De novo DNA (LaFleur, T. L., 2022)(Salis et al., 2009)(Figure 22). The results confirmed that Lrp expression was significantly below our expectations.

Figure 22. Translation rates with and without the RBS point mutation. A, Translation rate without the RBS point mutation. B, The decreased translation rate affected by the RBS point mutation.

Thus, we decide to fix this mutation.

Cycle #2: PpchA-RBS-Fixed biosensor validation

Design (Cycle #2)

We designed primers to fix the "T" of mutated RBS to "C" as the following.

Figure 23. Difference of the two plasmids.

See more details on design page.

Build (Cycle #2)

PCR and Golden Gate assembly were performed to achieve the change in sequence, fixing the wrong "T" back to "C". Agarose gel electrophoresis was performed and the bands indicate the correctness of the components' length.

Figure 24. The AGE result of the PCR product for the construction of plasmid PpchA-RBS-fixed. A, materials of the construction. B, Goldengate assembly result of plasmid construction. The band at 5278bp in (B) indicated the success in plasmid construction.

Test (Cycle #2)

The RBS-fixed version of the plasmid is tranformed into E.coli strain Trelief 5α, and performed the same kinetics analysis as the previous plasmid verification.

Overnight cultures of PpchA-RBS-fixed transformed Trelief 5α cells in LB media at 37°C were 1:250 (v/v) inoculated into fresh M9 media with glucose,casamino and 1 thiamine, cultured for 3 hours and diluted to OD=0.01 using M9 media. Sodium butyrate was added to gain final concentrations of 100mM, 50mM, 25mM, 12.5mM, 6.25mM, 3.125mM, 1,5mM, 0.75mM, 0.38mM and 0mM as control (see protocol). The detailed plate setup is shown below in Table 4.

The results of the kinetics assay were used to calculate the fluorescence/OD curve (Figure 25). The fluorescence/OD curve of the groups showed a more significant difference, and a dynamic range that is around twice of the previous result. This shows that our fix was successful, and the pPcha/Lrp system could sense butyrate to some extent.

Table 4. Plate layout of kinetics assay for PpchA-RBS-Fixed.

Figure 25. Kinetics of GFP expression over 16.7 hours. Fluorescence / ABS600 was used to represent GFP expression.  

Learn (Cycle #2)

We successfully integrated the pPcha/Lrp sensor system with the cI repressor system (NOT gate) together to create a functional SCFA sensor, which could induce the expression of the gene of interest that is inversely proportional to the concentration of SCFA. However, through this result, we found the dynamic range of the pPcha promoter to be relatively small, only around 2 folds, which would raise certain limitations in our future product design. Thus, we want to explore some different possibilities of sensing butyrates. We chose to characterise pHpdH/HpdR from Pseudomonas putida.

Reference

d’Oelsnitz, S., Stofel, S. K., Love, J. D., & Ellington, A. D. (2024). Snowprint: a predictive tool for genetic biosensor discovery. Communications Biology, 7(1). https://doi.org/10.1038/s42003-024-05849-8

Kineret Serebrinsky-Duek, Barra, M., Danino, T., & Garrido, D. (2023). Engineered Bacteria for Short-Chain-Fatty-Acid-Repressed Expression of Biotherapeutic Molecules. Microbiology Spectrum, 11(2). https://doi.org/10.1128/spectrum.00049-23

LaFleur, T. L., Hossain, A., & Salis, H. M. (2022). Automated model-predictive design of synthetic promoters to control transcriptional profiles in bacteria. Nature Communications, 13(1). https://doi.org/10.1038/s41467-022-32829-5

Ruegg, T. L., Pereira, J. H., Chen, J. C., DeGiovanni, A., Novichkov, P., Mutalik, V. K., Tomaleri, G. P., Singer, S. W., Hillson, N. J., Simmons, B. A., Adams, P. D., & Thelen, M. P. (2018). Jungle Express is a versatile repressor system for tight transcriptional control. Nature Communications, 9(1). https://doi.org/10.1038/s41467-018-05857-3

Salis, H. M., Mirsky, E. A. & Voigt, C. A. (2009). Automated design of synthetic ribosome binding sites to control protein expression. Nature Biotechnology, 27(10), 946–950. https://doi.org/10.1038/nbt.1568

Taketani, M., Zhang, J., Zhang, S., Triassi, A. J., Huang, Y.-J., Griffith, L. G., & Voigt, C. A. (2020). Author Correction: Genetic circuit design automation for the gut resident species Bacteroides thetaiotaomicron. Nature Biotechnology, 38(8), 1001–1001. https://doi.org/10.1038/s41587-020-0545-9