Overview
After carefully designing and planning the experiment, we initiated the experimental part of our project. We mainly focused on characterizing the threshold guard in different signal-capturing systems, the orthogonality and efficiency of the quorum sensing module, the final signaling decay process, the kill switch, and computational results. We then collected our data and analyzed it according to the Engineering cycle (See page Engineering Success) to improve our experiment.
Detailed information regarding how we implement it and the record kept in the wet lab can be found in the Experiments part and the Notebook part.
Original XylS Threshold Guard Switch
As described by Ángel and Víctor, the "Digitalizer module" they built has a clearly defined on-and-off status. As mentioned in Design, we used it as a threshold guard switch that only allowed a specific inducer of a specific concentration to open our promoter. To verify this switch's functionality and minimum threshold, we conducted a series of gradient concentration tests using classical inducers of the Xyls/Pm system: Benzoic acid and 3MBz.
- Inducer: Benzoic acid
Initially, a wide range of 2000 to 0µM benzoic acid was added to E. coli BL21(DE3) (initial OD value = 0.688) that had the switch sequence (BBa_K3202045). msfGFP fluorescence was measured by a synergy HTX microplate reader with excitation at 485 nm (±20 nm) and emission at 520 nm (±20 nm). Fluorescence was normalized to the OD after the background fluorescence value was subtracted from all RFU data. Continually Measuring every 2 minutes for 12 hours, we surprisingly found that the turning-on threshold of this lay at a low level—between 150 μM and 100 μM.
Overall threshold test of guard switch (benzoic acid)
Besides, the time that this switch needed to turn on was collected when the relative fluorescent units (RFU) reached above 200, providing strong evidence that when the concentration of benzoic acid was above 150μM can be regarded as a line separating the switch's on-and-off status. However, the response time might not be as promising as we thought. The minimum time our switch cost to reach the on-status was 74.33 minutes at 2000μM Benz. Acid.
Response time of guard switch (benzoic acid)
Being curious about what happened when benzoic acid concentration changed from 100μM to 150μM, we performed a detailed threshold test around this range (80μM~150μM). Using the same bacteria (initial OD value = 0.480) and following the same protocol, we obtained the detailed situation shown in the figure below. Although the absolute value of RFU might be different from the previous test due to the initial growth status, we still found it distinguishable between the on (above 100 μM) and off (below 100 μM) models of this switch.
Detailed threshold test of guard switch (benzoic acid)
- Inducer: 3MBz
In order to ensure the on-and-off status on our threshold guard switch is non-specific to benzoic acid, we used 3MBz as another inducer. The result also provided an on-and-off threshold line at 10μM and aligned with the work done by Ángel and Víctor.
Overall threshold test of guard switch (3MBz)
PobR Threshold Guard Switch
- Determination of the Threshold of Our Switch
We grew E. coli (Fast-T1) with a chimeric PobR Threshold Guard Switch (msfGFP version, BBa_K4619010). We measured the quantity of msfGFP expression over time at different concentrations of 4-HBA using a synergy HTX microplate reader with excitation at 485 nm (±20 nm) and emission at 520 nm (±20 nm). The initial OD600 is 0.9.
The diagram shows the changes in msfGFP fluorescence levels over time for varying concentrations of 4-HBA.
After removing the control group and dividing it by OD600, we obtained the following results from our analysis. We observed that concentrations above 0.25mM efficiently activate the switch, leading to a relative fluorescent unit (RFU) more significant than 140 after a certain period. Conversely, intensities below 140 are considered non-activated, as 140 represents the asymptote of the curve with a concentration of 0.25mM, and the proteins produced at that concentration are just enough.
An experiment with smaller concentrations of 4-HBA was conducted to determine the specific threshold of the PobR Threshold Guard Switch. The experiment involved varying concentrations of 4-HBA at 0.6mM, 0.3mM, 0.25mM, and 0.2mM. The results indicate that the expected putative threshold, as predicted in the model, is between 0.2mM and 0.3mM.
Detailed threshold test of the PobR Threshold Guard Switch.
We believe the detection threshold can be lowered by replacing msfGFP with luxI in the final work. LuxI produces VAI, activating the downstream Quorum sensing system as a signal amplifier.
- Time Needed to Trigger the Switch
In addition, We need to analyze how long it takes to trigger the switch successfully. We define it as open when the RFU level reaches 140.
The diagram shows the time needed to trigger the switch at different concentrations.
By analyzing the data in the figure, we can observe that a concentration ranging from 0.5mM to 10mM requires approximately 120 minutes. Despite this duration being slightly longer than our initial estimates, it should have minimal impact on the final products because only a tiny quantity of LuxI is required to initiate the quorum sensing system.
- Specificity of PobR Threshold Guard Switch
We confirmed that our PobR system is specific to 4-HBA by testing its response with other inducer, such as benzoic acid. To demonstrate this, we incubated E. coli with either 5 mM 4-HBA or 5 mM benzoic acid and measured the fluorescence intensity.
Responses of our PobR system to 4-HBA and Benzoic Acid.
The results align with the engineering article cited, demonstrating the superior specificity of our system towards 4-HBA compared to its analogs.
- Leakage Control Performance of Our Switch
We conducted an experiment to test the effectiveness of our threshold guard switch in controlling leaks. We incubated two groups of E. coli - one with the threshold guard switch and the other without it. We then measured the A.U. (A.U. = (Intensity of Fluorescence -Basal Fluorescence)/OD600) of msf-GFP emitted by both groups of E. coli at different time when there was no inducer added.
The A.U. of msf-GFP in two groups of E. coli (Note: A.U. = RFU/OD600, same after)
We observed a significant reduction in leakage with our Threshold Guard Switch. The following photo shows that we can even see the difference between these two groups with our naked eyes.
The fluorescence Photos of these two groups of E. coli.
These results indicate that most of the leakage and false positive phenomena can be eliminated with the help of our threshold guard switch.
Orthogonal Quorum Sensing
Here, we designed AHL-induced quorum sensing systems without a positive feedback system.
We conducted an experiment to compare the efficiency of our quorum sensing systems with and without a positive feedback circuit. We achieved this by measuring the change in fluorescence values at different concentrations. First, we cultured E. coli containing plasmid BBa_K4619012 in LB medium until their OD600 reached 0.5-1.0. We then induced them with AHL (VAI) dissolved in DMSO medium and added 200 µl aliquots of each sample to a 96-well plate for 3 hours. Finally, we recorded fluorescence spectra using a fluorescence spectrometer at 485 and 528 nm for excitation and emission wavelengths, respectively. The results are as follows.
Quorum sensing efficiency test. Note: QS1-Circuit with luxI , QS2: Circuit without luxI.
As shown above, there are no significant differences between these two quorum sensing systems. Therefore, we finally chose the shorter circuit, non-positive feedback quorum sensing system. After that, we tested the Signal-noise performance of our quorum sensing system.
Signal-noise ratio vs. Time of our quorum sensing system.
As shown in the figure, at a concentration of 100 μM, a particular induction effect has been produced.
In order to test the orthogonality of our quorum sensing system, we compared the colonies' relative fluorescence intensity and OD value after adding the correct AHL (VAI) and the incorrect AHL (PAI).
Orthogonality test between VAI and PAI.
As can be seen from the above, after four hours of induction, the induction effect of adding AHL that matches the system is better than that of adding mismatched AHL, indicating that our system is not susceptible to interference and the LuxR has specificity;
Besides, we also observed the quorum sensing effect under the monitor of a fluorescence microscope.
The quorum sensing effect under the monitor of the fluorescence microscope.
Nanoluc
We separate Nanoluc into one plasmid to investigate the signal process. Nanoluc is constitutively expressed in the plasmid under the control of IPTG. When induced by IPTG, we perform luminescence measurement with the Nano-Glo® Live Cell Assay System from Promega. We successfully characterize the signal-time curve, gaining essential information about signal recession and other features. Moreover, we build models upon generated data, thus gaining a more thorough comprehension of the signaling process and how we characterize it.
Characterization of Luminescence (RLU)
After being induced by IPTG for 4h, we followed the Nano-Glo® Live Cell Assay System standard protocol for measuring luminescence, continuously recording for 2h.
We acquired the following data:
The result of the luminescence-time span plot
Compared with the negative control group, the group with furimazine added demonstrates a significant rise in luminescence level, which can be captured by our hardware undoubtedly. We further mix the substrate with E. coli that does not express nanoluc to ensure the positivity of former results.
Moreover, the decay of the signal follows a relatively predictable pattern, which is beneficial for modeling and overall characterization. More information regarding the model, please click here.
Kill Switch
Results of the Toxin-Antitoxin System
To examine the effects of our toxin-antitoxin system, we constructed a plasmid containing both the antitoxin gene (phD) induced by IPTG and the frequently expressed toxin gene (doc). We plan to culture engineered bacteria carrying suicide switch plasmids on two solid media, one without IPTG and the other containing 0.1M IPTG, representing the experimental and positive control groups, respectively. Since the survival of the bacterial strain needs to be ensured during plasmid transformation, the use of a liquid medium containing IPTG is inevitable. Therefore, to make the experimental results more rigorous, we performed multiple rounds of washing with deionized water on a portion of the experimental group of engineered bacteria. Thus, the experimental group was divided into two subgroups: one with trace amounts of IPTG residue (→0) and one completely free of IPTG (0). As shown in the figure below, it can be seen that the engineered bacteria can survive when there is a certain concentration of IPTG in the culture environment. However, without IPTG, the engineered bacteria have a significantly lower survival rate. Comparing the group with inducer residue concentration approaching 0 and the group equal to 0, the survival status of the engineered bacteria also conforms to the experimental expectation.
There are some problems with these experimental results: The engineered bacteria that still survive in the environment without IPTG, the reason should be that the lac promoter used in the gene circuit has a certain leakage expression problem. Also, the number of engineered bacteria on the solid medium containing the inducer is not very large. The reason for this should be that the J23119 promoter controlling the expression of the toxin gene on the vector is too strong compared to the lac promoter upstream of the antitoxin. The above problems can be better balanced by replacing the promoter pairs with different initiation and leakage strengths and using toxin-antitoxin pairs with different antagonistic strengths.
The Verification of the Toxin-Antitoxin Kill Switch
Rational Design
According to the structural database provided by UniProt and results from Alphafold, we first collected the structures of LuxR and its homologs in other species, including SdiA from E. coli and LuxR from Yersinia enterocolitica. Then, we perform molecular docking on the interaction between VAI and the wildtype LuxR from Vibrio fischeri. We aim to explore the binding pattern between VAI and LuxR, therefore providing insights into further engineering the binding process to be more specific and efficient.
The Result of the Alignment
By aligning the architecture of three structures of LuxR, we distinguished the similar binding pocket of LuxR-VAI, and some amino acid residues nearby play important roles. Moreover, the position of VAI in LuxR remains similar, indicating core polar interactions.
The docking results between VAI and two LuxR homologs and the target LuxR(last docking position)
Hydrogen bonds are formed between the nitrogen from amide group and hydroxyl group from Serine or between structured water and the hydroxyl group from Serine.
Compared with the predicted structure, the original serine is replaced by leucine, which eliminated the hydrogen bond. Moreover, the orginal valine is replaced by arginine, which partially compensate for the polar interaction. We then concluded the possible rational design to be two mutations based on understanding the interaction between AHL and the protein, including Leu 42 to Ser and Arg 67 to Val.
