Here you can find the results of our experiments. Details of protocol adjustments can be found in our notebook.
Since we cloned 20 constructs, we noticed early on how important it is to agree on names for the constructs that are preferably short and easy to understand. To make it easier for you to understand our results, we have created Table 1 that gives an overview about the name and the construct details.
These constructs were cloned and cloning success was verified by sequencing:
Table 1: Construct Name Explanations. This table shows the construct abbreviations we used during our entire project as well as the riboswitch construct and reporter gene combination it consists of. Abbreviations: T7 Prom: T7 Promoter, SR_1: Spacer Region 1, Li+-II_RS: Li+-II Riboswitch, PRBS: Predicted RBS, nhaA: Beginning of nhaA gene. For more detailed explanations on the constructs see project description.
After through testing, we found out that only the R1 and R5 constructs produced reliable results (ie. 1G, 5G, …). We therefore submitted the following parts:
Basic parts: BBa_K4654000 (sfGFP), BBa_K4654001 (mScarlet-I3), BBa_K4654002 (nanoLuc), BBa_K4654003 (lacZ), BBa_K4654009 (Spacer1), BBa_K4654010 (Spacer2), BBa_K4654011 (Li+II-Riboswitch), BBa_K4654012 (5’ nhaA-gene), BBa_K4654013 (T7 Promoter)
New Best Composite Part: BBa_K4654018 (1N)
Composite Parts: BBa_K4654021 (1G), BBa_K4654021 (5G), BBa_K4654022 (5M), BBa_K4654023 (5N)
In the initial stages of our research, we conducted cell-based assays using LB low sodium media with a pH set at 9. This specific pH level was chosen because it is known that bacteria tend to import more lithium at higher pH levels. However, we quickly observed that this had a significant negative impact on bacterial growth in liquid culture.
In response to this issue, we decided to explore how varying the choice of media and adjusting the pH might influence both cell growth and the expression of our reporter gene. To assess this, we utilized a genetic construct in which the expression of sfGFP was controlled by the rhamnose-inducible T7 promoter. This is the construct that we usually used as positive control in our riboswitch reporter assays with added lithium.
We induced liquid E. coli KRX cultures that carried this genetic with 0.1% rhamnose. We then measured optical density (OD600) and sfGFP expression levels using a Tecan Spark plate reader. These measurements were taken at specific time points, starting 30 minutes after rhamnose induction and continuing at 30 minute intervals for a total of 11 readings.
Figure 1: E. coli KRX 5G Growth in Different Media. The figure shows OD600 values for E. coli KRX 5G bacteria over a timeframe of 330 Minutes. Measurings were done every 30 minutes with a Tecan Spark Plate reader. Bacteria were grown overnight in respective media as pre cultures. The next day, main cultures were inoculated and grown until OD600=0.1, then 0.1% rhamnose was added to induce sfGFP expression.
Figure 2: sfGFP Expression in Different Media. The figure shows sfGFP expression for E. coli KRX bacteria transformed with the 5G construct over a timeframe of 330 Minutes. Bacteria were grown overnight in respective media as pre cultures. The next day, main cultures were inoculated and grown until OD600 0.1, then 0.1% rhamnose was added to induce sfGFP expression. sfGFP expression was measured with a Tecan Spark Plate Reader. Normalization of sfGFP expression values were done by dividing the RFU by the respective OD600 values.
Notably, we observed that while growth in LB low sodium media was the slowest, the expression of sfGFP reached its highest values. This might be because the salt level in the media affect the measurement, which would also explain why the sfGFP expression in bacteria grown in TB medium is low even though they have the best nutrient concentrations in the medium.
Figure 3: E. coli KRX 1M and 5M Growth Under Different pH Conditions. The figure shows OD600 values for E. coli KRX 1M 5M bacteria growth in LB media at different pH. Measurements were done with a Tecan Spark Plate reader, the settings for the plate reader are in the protocol "sfGFP / mScarlet-I3 Assay". Bacteria were grown overnight in respective media as pre cultures. The next day, main cultures were inoculated and grown until OD600=0.1, then 0.1% rhamnose was added to induce mScarlet-I3 expression.
Figure 4: Response of Riboswitch-mScarlet-I3 Constructs to pH 7.5, 8, 8.5. The figure shows RFU values divided by the respective OD600 values. The 1M and 5M constructs were used. Measurement was done with Tecan Spark Plate reader, the settings for the plate reader are in the protocol "sfGFP / mScarlet-I3 Assay". Bacteria were grown overnight as pre cultures. The next day, main cultures were inoculated and grown until OD600 0.1, then 0.1% rhamnose and 50 mM LiCl was added to the cultures to induce mScarlet-I3 expression.
We observed that an increase in pH inhibited both bacterial growth and reporter expression. Therefore, we have decided to use pH-neutral media from now on.
We were very excited that our cloning worked so well because this meant that we could put our entire focus into finding the most suitable construct for our test system. Because of the amount of different riboswitch-reporter combinations, the wet lab team decided to do separate assays with the different reporters and met on a regular basis to compare the results. This enabled us to find the best system in the cell-based environment and discard some combinations before we started the very -expensive cell-free assays, saving us time and money. This is why you will first find the individual group results and a summary and comparison in the end.
1G can be identified as the most suitable construct, since the riboswitch can close and open depending on the LiCl concentration. The induced expression is significantly higher than the uninduced expression. At the same time, the constructs 2G, 3G and 4G can be excluded for our test system, since the expressions differ significantly less. The riboswitches can not close and open depending on the LiCl concentration. 4G can instead be used as a positive control for uninduced expression. Furthermore, 5G can be confirmed in its function as a positive control, since the induced expression clearly exceeds that of the other constructs.
Figure 5: Response of Riboswitch-sfGFP Constructs to 50 mM LiCl. The figure shows RFU values divided by the respective OD600 values. The 1G, 2G, 3G, 4G and 5G constructs were used. Measurement was done with Tecan Spark Plate reader, the settings for the plate reader are in the protocol "sfGFP / mScarlet-I3 Assay". Bacteria were grown overnight as pre cultures. The next day, main cultures were inoculated and grown until OD600=0.1, then 0.1% rhamnose and 50 mM LiCl was added to the “induced”-cultures to induce sfGFP expression.
The sfGFP expression of the cell culture with a 0 mM LiCl induction is strongly significantly different (**, p=0.001) from the sfGFP expression of the cell culture with a 1.5 mM LiCl induction (t=-4.031;alpha=0.05;df=16;crit.t=2.12,p=0.001). Therefore, the therapeutic range of LiCl treatment (< 1.5 mM LiCl) can be reliably covered.
Figure 6: Response of Riboswitch-sfGFP Construct 1G to 0 and 1.5 mM LiCl. The figure shows RFU values divided by the respective OD600 values. The 1G construct was used. Measurement was done with Tecan Spark Plate reader, the settings for the plate reader are in the protocol "sfGFP / mScarlet-I3 Assay". Bacteria were grown overnight as pre cultures. The next day, main cultures were inoculated and grown until OD600=0.1, then 0.1% rhamnose and 0 - 1.5 mM LiCl was added to induce sfGFP expression.
The sfGFP expression of the cell culture with a 0.5 mM LiCl induction is significantly different (*, p=0.03) from the sfGFP expression of the cell culture with a 1.5 mM LiCl induction (t=-2.1039;alpha=0.05;df=68;crit.t=1.99,p=0.03). Therefore, a reliable distinction can be made between a non-toxic concentration (0.5 mM LiCl) and a toxic concentration (1.5 mM LiCl). This proves the suitability of the 1G construct for our test system.
Figure 7: Response of Riboswitch-sfGFP Construct 1G to 0.5 and 1.5 mM LiCl. The figure shows RFU values divided by the respective OD600 values. The 1G construct was used. Measurement was done with Tecan Spark Plate reader, the settings for the plate reader are in the protocol "sfGFP / mScarlet-I3 Assay". Bacteria were grown overnight as pre cultures. The next day, main cultures were inoculated and grown until OD600 0.1, then 0.1% rhamnose and 0.5 - 1.5 mM LiCl was added to induce sfGFP expression.
Since the 1M construct cannot reliably discriminate between different LiCl concentrations, we decided not to consider mScarlet-I3 for our experiments with cell-free systems. We suspect that the mScarlet-I3 sequence interferes with the proper folding of the riboswitch.
Figure 1: Response of Riboswitch-mScarlet-I3 Construct 1M to 0.2, 0.5, 1, 1.5, 50 mM LiCl. The figure shows RFU values divided by the respective OD600 values. The 1M construct was used. Measurement was done with Tecan Spark Plate reader, the settings for the plate reader are in the protocol "sfGFP / mScarlet-I3 Assay". Bacteria were grown overnight as pre cultures. The next day, main cultures were inoculated and grown until OD600 0.1, then 0.1% rhamnose and 0.2 - 50 mM LiCl was added to induce mScarlet-I3 expression.
Because the signal strength of NanoLuc does not only depend on the amount of enzyme, but also the substrate concentration (Furimazine), we decided to measure the signal development under different conditions. Verifying the signal half-time allows us to avoid exceeding the measurable signal length due to high enzyme concentration and to verify that our experimental conditions do not disrupt the catalysis of the bioluminescence reaction by NanoLuc.
We decided to compare the signal development for different dilutions of two bacterial E. coli strains: KRX and BL21(D3). The cultures were diluted before application of the NanoLuc assay and three technical replicates were measured. The experiment was conducted with the 5R riboswitch construct to avoid influences caused by the lithium concentration. The measurements were performed with two different attenuation methods. For method “OD1”, the signal intensity is attenuated by the factor 10, while the attenuation method “none” applies no attenuation filter.
All dilutions show an increase in signal intensity until 30 minutes. The induction only has a significant influence for dilution 1:100. With dilution “1:1000”, the measured signal values show a higher error. Because of the results we hypothesize that for dilutions higher than “1:100” the NanoLuc concentration is too low and lower dilutions are more suitable under our experimental conditions. The dilution “1:100” confers with the expected half-time of 120 minutes (Figure 1).
Figure 1: Signal development over time and influence of induction in E. coli KRX for different dilutions of cell culture. The graphs compare the measurement values between inducted and not inducted cultures. The cultures were diluted before application of NanoLuc assay, the negative control was measured without assay. The columns show different attenuation methods and signals exceeding the range were set to 0. Three technical replicates were measured. The time represents the time passed after the incubation time of the diluted cultures and the NanoLuc assay was finished.
The measurements for BL21(D3) showed a higher signal intensity than for KRX. The dilution “1:100” exceeded the measurable signal strength for the measurement method without attenuation filter. For dilution “1:100”, the signal value increases until 60 minutes and starts to decrease after 95 minutes. For the higher dilutions, the signal increase persists for 120 minutes and only decreases slowly. For all dilutions, the induction has a significant influence. To confirm the dilutions confer with the expected halt life of 120 minutes, the measurement had to be performed for a longer time (Figure 2).
Figure 2: Signal development over time and influence of induction in E. coli BL21(D3) for different dilutions of cell culture. The graphs compare the measurement values between inducted and not inducted cultures. The cultures were diluted before application of NanoLuc assay, the negative control was measured without assay. The columns show different attenuation methods and signals exceeding the range were set to 0. Three technical replicates were measured. The time represents the time passed after the incubation time of the diluted cultures and the NanoLuc assay was finished.
Of the riboswitches 1N, 2N and 3N, only the riboswitch 2N reacted on lithium in E. coli BL21(D3) with a significantly higher signal intensity. In KRX, only 1N shows a significantly higher signal intensity, while 2N shows a slightly higher signal with lithium. Because the measured signals for the constructs that did not react to lithium are very low, we conclude that these riboswitches did not unfold in response to lithium and the measured signals represent the base signals of the unfolded riboswitches (Figure 3).
To examine the reaction of our riboswitch constructs to lithium, we tested all riboswitch constructs with a concentration of 10 mM lithium. Because of the results in the previous experiment, the dilution was decided based on the signal intensity. For E. coli KRX, the cultures were diluted 1:100 before the NanoLuc assay was applied. For E. coli BL21(D3), the cultures were diluted 1:1000 before the NanoLuc assay was applied.
Figure 3: Reactions of the riboswitch constructs 1N, 2N, and 3N on lithium in E. coli KRX and BL21(DE3) The graphs compare the measurement values between cultures incubated with and without 10 mM lithium. The cultures were diluted before application of NanoLuc assay, the negative control was measured without assay. Three technical replicates were measured with the attenuation method “none”. The time represents the time passed after the incubation time of the diluted cultures and the NanoLuc assay was finished.
The measured signal of the riboswitch R4 is lower for the samples treated with lithium than for the samples without lithium in both strains. Because the measured signals for riboswitch R4 are relatively high in comparison to R1, R2, and R3 in both strains, this may indicate that riboswitch R4 is also unfolded in absence of lithium (Figure 4).
Figure 4: Reactions of the different riboswitch construct R4 on lithium in E. coli KRX and BL21(DE3) The graphs compare the measurement values between cultures incubated with and without lithium. The cultures were diluted before application of NanoLuc assay, the negative control was measured without assay. Three technical replicates were measured with the attenuation method “none”. The time represents the time passed after the incubation time of the diluted cultures and the NanoLuc assay was finished.
The lacZ sequence was too long to be synthesized in one DNA fragment, which is why we had to divide the sequence into fragments and subsequently assemble them using the Golden-Gate-Assembly (GGA) method.
In the initial assembly attempts, we encountered problems with missing fragments, which were later identified through sequencing. This highlighted the need for careful quality control and validation in our process. Once the cloning was successful, we were able to succeed with the reporter assays:
Our first reporter system involved a relatively simple X-Gal liquid culture assay setup. We were initially excited about the potential of our basic idea, and indeed, it showed promise. However, we soon realized that the X-Gal assay yielded binary results rather than providing quantitative measurements.
The cell-based experiments, particularly those involving the ONPG assay, proved to be more complex than anticipated. One significant challenge was the high toxicity of certain chemicals, primarily beta-mercaptoethanol and chloroform, which necessitated the use of a fume hood for safety reasons. To measure the reporter expression, we used a photometer to detect the wavelengths 420nm, 550nm, and 600nm.
All cell-based lacZ/beta-galactosidase assays were carried out in E. coli KRX bacteria as they do not carry a lacZ gene in their genome.
For the ONPG Assay setup, using a 10mM LiCl concentration. Notably, we observed the highest level of activity in the 4L construct. To enhance the reliability of our findings, we incorporated technical replicates into the experimental design. All results were tested against a negative control.
However, conducting these experiments posed certain challenges. The fume hood setup and the growth conditions of the bacterial cultures hindered our ability to perform high-throughput measurements, as each sample had to be individually measured in a cuvette.
Building upon this initial procedure, we conducted a subsequent series of experiments using different LiCl concentrations (0.5mM, 1.0mM, and 1.5mM) where we detected a slight decrease in OD420. This decline could be attributed to either the inherent imprecision of the borrowed photometer or the possibility of disintegration within the coloring ring systems of the molecules. Consequently, we approached the data with a high degree of skepticism, recognizing the potential impact of these variables on our results.
Additionally, it's worth noting that E. coli KRX strains carrying the lacZ plasmid exhibited a reduced growth rate when compared to the other constructs, introducing an additional layer of complexity into our experiments.
In light of these challenges, we recognized the importance of addressing technical issues, such as variations in sample handling, assay timing, and equipment calibration. These technical challenges had the potential to introduce variability and confounding factors into our results, making the accurate interpretation of the data a significant concern.
Figure 1: ONPG assay comparing the reactions of the constructs R1, R2, R3, and R4 on lithium.
Given the challenges of the ONPG assay, we made the decision to shift our focus towards utilizing reporters that were easier to handle and more reliable. This shift was motivated by our commitment to maintaining very high standards for precision and accuracy in our experiments. Additionally, we sought reporters that were less toxic, mitigating potential safety concerns associated with certain chemical agents.
Here you can see the riboswitch-lacZ assay results summarized in bullet points:
For our cell-free assays, we used the biotechrabbit RTS E. coli HY KIT for cell-free protein expression. The experiments were conducted according to the manufacturer's instructions. We tested the 1G and 1N construct. The results show clearly that our test system can detect lithium ions in a cell free system but needs some adjustments regarding the quantification.
Figure 5: Construct 1G in Cell-Free Assay. Response of Riboswitch-sfGFP Construct 1G to 50 mM LiCl compared to an uninduced sample as negative control in a cell-free environment. The figure shows RFU values, the measurement was done with Tecan Spark Plate reader, the settings for the plate reader are in the protocol "sfGFP / mScarlet-I3 Assay". For cell-free protein expression we used the biotechrabbit RTS E. coli HY KIT. The experiments were conducted according to the manufacturer's instructions. 50mM LiCl was added to induce sfGFP expression.
Figure 5 shows that after the addition of 50 mM LiCl, construct 1G produces a high luminescence signal in the cell-free assay. However, in this experiment we did not test how the system responds to different LiCl concentrations which is something that would need to be considered in further experiments.
Figure 6: 1N Luminescence Expression in a Cell Free Environment. Response of Riboswitch-NanoLuc Construct 1N to 0.5, 1, 1.5 and 5 mM LiCl compared to 0 mM LiCl as an uninduced sample as negative control in a cell-free environment. The figure shows counts per second values, the measurement was done with Tecan Spark Plate reader, the settings for the plate reader are in the protocol "NanoLuc Assay". For cell-free protein expression we used the biotechrabbit RTS E. coli HY KIT. The experiments were conducted according to the manufacturer's instructions. LiCl was added to induce NanoLuc expression. The obvious increase of signal between 0 mM and the reactions with LiCl show that the construct also works in a cell-free environment. However, it is also clearly visible that the quantification did not work as one would expect the signal to get stronger with a higher LiCl concentration. This led us to the conclusion that we need to repeat the experiment with replicates. The measurements were done 10 minutes after addition of the substrate to the reactions. Measurements were done with a Tecan Spark plate reader.
Figure 6 shows very nicely that the 1N construct shows increased reporter expression in the presence of lithium compared to an uninduced control. However, contrary to expectations the signal decreases with increased LiCl concentration. No replicates were conducted in this experiment, which could be the reason why we did not get reliable results for this experiment. Another explanation could be that due to the higher activity of the reporter under increased LiCl concentrations, the substrate is used up faster which could lead to decreased luminescence.