Proof of concept: Favourable growth environment for E.coli at high pH (pH ~6.5).
Design using compuational predictions: Design and testing of nine sgRNA for flux redirection into increased lactate production.
Modular engineering: Building a level 2 sgRNA part to knock down of multiple enzymes simultaneously, leading to a constant and efficient flux redirection towards lactate.
Verification: Successful production and secretion of C4-HSL and confirmation by FIA-MS analysis.
Improved documentation: Further characterisation of the LasB promoter by induction via RhlI and RhlR.
Implementation: Integration of a new lactate dehydrogenase enzyme into E. coli which produces favourable L-lactate isomers promoting vaginal health.
(Preparations for) genomic integration: Design and amplification of a homology arm pair to knock out the ackA gene.
Although extensive effort was made to reproduce a functional vaginal fluid simulant (VFS), as described in literature, no satisfying results were obtained. Mostly the high pH condition (pH ~6.5) were more likely to show growth of both L. crispatus and E. coli, however, the low pH conditions (pH ~4.5) never worked for both bacteria simultaneously.
We performed growth curves measuring liquid culture OD at 600nm, as well as counted CFU for the purpose of plotting growth curves. The CFU protocol did not yield a representative growth curve. In the future, more optimization in this protocol needs to be done, such as adjusting the dilution factor, the media for plating out the bacteria and increasing the measured time intervals to gain more data points.
We created growth curves in LAPTg medium, a media specifically designed for cultivating Lactobacillus bacteria. Even at pH at ~6.5, L. crispatus outperformed E. coli, as shown in Figure 1. We kept these results for the sake of replicating them, if we were able to successfully finish our project and co-culture both bacteria in the end. Obtaining co-cultures was also attempted, but did not lead to any useful results since the bacterial colonies were not distinguishable on the agar plates. Improvements for the future would include more screenings for selective media or genetic manipulation by inserting fluorescent labels to distinguish bacteria by color.
Figure 1: Comparison of the growth behaviour between E. coli (left) and L. crispatus (right) in LAPTg.
Eventually we decided on a VFS medium that allows us to illustrate the growth behaviour in dysbiotic conditions, shown in Figure 2. The liquid OD 600nm measurements were successful and nicely confirmed one aspect of our concept: If we were to use E. coli as a probiotic target, it will be able to grow better at dysbiotic conditions, such as high pH. This allows us to utilise the engineered quorum sensing system and induce conditional lactate production, as shown further down.
Figure 2: The lines represent the growth of E. coli and L. crispatus in the specific condition. The dark colors represent the high pH condition, the light colors represent the low pH condition.
Both bacteria show growth at high pH of the medium, where E. coli clearly outcompetes L. crispatus, supporting the use as a probiotic for dysbiotic conditions. Despite literature findings, we could not confirm that L. crispatus favours low pH conditions, this will be part of future research to screen for more suitable media representing the vaginal environment.
We designed nine different guide RNAs to knock down genes of interest that also use pyruvate as a substrate, and thus re-directed the flux of glycolysis to increase lactate production.
As mentioned in the engineering part, our initial single guide RNA design did not work out due to the fact that our PCR fragments were too short to perform the PCR clean up with the spin columns.
We tested the ethanol purification and unfortunately only one out of nine assembly was confirmed by a positive sequencing result.
The rest contained many mutations or either PCR or assembly didn't work out.
Confirmed sequencing of our designed plasmid after PCR and ethanol purification:
Therefore, we redesigned, reordered and reassembled our multiple single guide RNAs with great success.
The sgRNAs were expressed constitutively, while the dCAS9 enzyme was expressed on a tetracycline inducible promoter. For the measurement of D-lactate, we used a kit that allows us to measure the extinction coefficient of NADH at 340 nm over time and thus we could quantify the lactate concentration in our sample. The standard curve, as shown in Figure 3, has a high R2 value, showing high accuracy of the kit.
Figure 3: Standard curve for D-lactate measurement
The lactate production of each transformed bacterium was measured in two conditions: "uninduced" as a baseline how much lactate is produced normally and “induced” by 5 ug/ml tetracycline.
Figure 4a shows nicely that the expression of each guide RNA leads to an increase in lactate production.
It is evident that the lactate concentration drops in the 19.5h timestep.
We hypothesise that upon starvation of the bacteria, the nutrients are not enough to support growth anymore, so the bacteria start to catabolize the lactate within the medium.
There was only a 10% difference between efficiency predicted at the design stage, and real efficiency.
With our results, we are neither able to support nor deny the same efficiency order as the predictions.
Figure 4a. Subplots A-I show lactate concentration changes after the insertion of a level 1 sgRNA part targeting pflA (A-C), pflB (D-F) or adhE (G-I) in the engineered E. coli, in the induced versus uninduced state.
Finally, we wanted to maximise the lactate production by creating a level 2 vector, containing 4 different sgRNAs to efficiently target every protein of interest.
Based on the results obtained from the individual sgRNA, targeting area and predicted efficiency, we decided to assemble pflA1, pflB1, adhE1 and adhE2.
Surprisingly, the effect observed in Figure 4b was less prominent than we had expected. The standard deviations are very small, which supports the idea that knocking down multiple enzymes simultaneously leads to a more consistent flux redirection towards lactate.
Figure 4b. Assembly of pflA1, pflB1, adhE1 and adhE2 sgRNAs as a level 2 part which knocks down multiple genes simultaneously. The standard deviations of measurements at every timestep are reduced compared to those seen in the respective sublots of constituent level 1 parts in Figure 4a.
We conclude that the sgRNA design was successful and can be used as a reliable tool to manipulate the central metabolism in the future. The combination of the guide RNAs leads to more available pyruvate in the cell, which future users can use to redirect the flux of e.g. glycolysis into a different desired direction.
Characterising the quorum sensing system was a vital part of our project, as it was our fundamental approach to having an inducible system.
Since the vaginal environment for the bacteria is highly dynamic and still poorly characterised, we steered away from an inducible system by e.g. pH.
Quorum sensing is based on population density and thus allows a robust control of the switch-like activation of its downstream transcription.
Originally found in Pseudomonas aeruginosa, we implemented the “Rhl quorum sensing” into our engineered bacterium.
The quorum sensing requires two parts, the production and the sensing unit.
RhlI is a protein that produces a small molecule called N-butanoyl-L-homoserine lactone (C4-HSL). This small molecule acts as a signalling molecule and has been described to be well diffusible (1,3). Using flow injection analysis mass spectrometry (FIA-MS), we were able to detect our molecule of interest in the transformed bacteria. Although we could not quantify the concentration of the molecule, we were able to see a clear difference between the negative control to our engineered E. coli. In Figure 5, the data analysis from the FIA-MS measurement is shown.
Figure 5: FIA-MS analysis results at 170 m/z, which corresponds to the deprotonated C4-HSL molecule. A3 = supernatant of untransformed E. coli, B3 = supernatant of engineered E. coli, C1 = extracted cell pellet of untransformed bacterium, D1 = extracted cell pellet of engineered E. coli
The fact that some signal in the negative control is visible, might be due to the fact that some bigger molecules fragmented during mass spectrometry, or that dimerization of two smaller molecules resulted in a peak at the same molecular mass. This data shows clearly that the molecule of interest is produced and secreted successfully.
Confirmed sequencing of our designed plasmid:
Now that we confirmed the expression of a functional RhlI enzyme, it is required to be sensed as well. RhlR is an enzyme that acts as a transcription factor, as soon as the concentration of C4-HSL reaches a certain threshold and can bind to RhlR. In the bound state, RhlR gets active and induces transcription of specific genes. Traditionally, the Rhl quorum sensing system uses the RhlAB promoter. However, this part has already been characterised in iGEM. We wanted to increase the iGEM parts collection by confirming that RhlR is also able to induce the LasB promoter (2). This is the consequence of a crosstalk between two different quorum sensing systems. This finding is useful for future teams, since this can be used to finetune the expression of the protein of interest dependent on the quorum sensing . However, the strength of the promoters needs to be compared first.
For testing the RhlR ability to bind to the LasB promoter, we used a level 2 plasmid which contains the RhlR and LasB expressing CFP when induced.
We performed a 24h time incubation and measured the fluorescence of CFP over time, to check whether the intensity increases upon adding a higher amount of C4-HSL manually to the medium of the bacteria.
The untransformed bacteria acted as a negative control.
As shown in Figure 6a, the fluorescence already increases 2 hours after induction of the system.
During the exponential phase, the bacteria are metabolically very active and thus the promoter might be leaking some expression of CFP, as we see in the hours 0-16.
After the cells reach saturation, we don't observe a stronger induction of fluorescence at e.g. 1 mg/ml C4-HSL compared to 0.01 mg/ml, since the quorum sensing system is a switch like mechanism, resulting in either complete induction or no induction at all.
The system is very sensitive, becoming completely induced at 0.01 ug/ml C4-HSL.
Figure 6a. Raw measurement of fluorescence. The untransformed E. coli also has a fluorescent signal at 429 nm excitation.
Figure 6b. Corrected values for the fluorescent signal of the negative control.
Confirmed sequencing of our designed plasmid:
The final step for complete autoinduction required a co-transformation to clone level 2 RhlR and LasB with CFP with the Level 1 RhlI plasmid into one bacterium. With the CFP output, we measured the fluorescent signal over time to confirm that the CFP signal increases as soon as the population density of the bacteria reaches a threshold density, which seems to be at OD 0.5. The growth rate of the engineered bacterium is slower, as it reaches the log phase around 2h later than its wild-type counterpart. In Figure 7, we can see that this resulted in complete characterization of our Rhl-Las quorum sensing hybrid system and is easily adapted by exchanging the CFP moiety to the protein of interest. Despite the slower growth rate and a ~30% lower stationary population density than observed in the wild-type, the engineered bacteria outputs nearly double of the fluorescence of the wild-type bacterium.
Figure 7. CFP output of wild-type vs. induced engineered E. coli MG1655 (left), as well as simultaneously measured OD values of the two populations (right). At the 6 hour mark, the fluorescence of engineered E. coli overtakes that of the uninduced one due to the production of CFP at a threshold density of OD ~0.5.
Confirmed sequencing of our designed plasmid:
To increase lactate production in E. coli, we inserted an additional lactate dehydrogenase enzyme, which originates from Streptococcus bovis and is a L-lactate dehydrogenase. There were several reason why we chose this enzyme: first, it has already been successfully overexpressed in E. coli; second, it produces L-lactate not naturally present in high quantities in E. coli; third, the L-lactate may be more favourable for the vaginal microbiome (4). To test the efficiency of our L-LDH, we used a L-lactate kit that gave luminescence readouts of luciferin, which was measured and used to quantify the amount of L-lactate in the sample.
As a negative control, we were not only able to compare with untransformed bacteria, but also an LDH knockout strain obtained from the KEIO knockout library. We used this strain to express our LDH under a constitutive promoter. Additionally, we characterised the lactate production dependency on the LasB promoter, which can be seen in Figure 8. Surprisingly, the constitutive promoter did not yield high amounts of lactate.
As a next step, we would have exchanged the promoter for a stronger unit and repeated the experiment. As expected, the induction of the LasB promoter yielded good results, as this had already been characterised with the CFP expression beforehand.
Figure 8. Measurement of L-lactate abundance in control strains; untransformed and knock out. Comparison to induced L-LDH via LasB promoter.
Confirmed sequencing of our designed plasmid:
For the maximal flux redirection, we were planning on knocking two enzymes out through chromosomal integration. We successfully performed PCR to amplify the one homology arms pair out of the bacterial genome.
Confirmed sequencing of our designed plasmid:
For chromosomal integration, we would have needed the pJUMP47ts plasmid, which we were unable to transform. Combined with time constraints, we were not able to complete this step.
We managed to...