Our LactoBack helper bacterium combines a higher-efficiency lactate dehydrogenase (LDH), CRISPRi-mediated flux redirection as well as population density sensing using the quorum sensing (QS) system found in P. aeruginosa. To test our system, we performed Design-Build-Test-Learn cycles on four system components: QS, LDH expression, dCas9 expression and sgRNA expression.
All our Designed Parts and Successful Transformation and assembly can be found here.
We aimed to engineer a bacteria that would restore a “healthy” Lactobacillus-dominated vaginal microbiome by secreting more lactic acid which exhibits [1.].
Figure 1: Schematic overview of the helper bacterium system circuit.
A quorum sensing mechanism is dependent on population density and leads to a conditional activation of dCAS9 expression and lactate dehydrogenase expression.
This results in flux redirection towards lactate production.
Figure 1 is an illustration of the designed system to elevate lactic acid production in our "helper bacterium".
The overall amount of lactic acid produced is dependent both on the amount produced per bacterium and the number of bacteria.
To maximize production of lactic acid in a single bacterium we planned to overexpress lactic acid dehydrogenase (LDH) and redirect using and a multiplexing construct containing four .
However both these methods lower the fitness of the helper bacterium, making it susceptible to being outcompeted by WT bacteria.
We therefore decided to couple the overproduction of lactic acid to a [2.] system.
This enables the bacterial population to grow fast initially, until it reaches a threshold level where expression of additional LDH and dCas9 are induced.
Moreover it maximizes the long term output of lactic acid by ensuring the population is of sufficient size to produce a substantial amount of lactic acid.
The components of this system were assembled in four separate plasmids that could later be chromosomally integrated (Figure 2).
Because some of the system's components (dCas9, sgRNAs, see following sections), require the screening of different variations to perform optimally, we decided to assemble our plasmids using [3.] and the Joint universal modular plasmids (JUMP) collection [4.].
Figure 2. Plasmid maps of LactoBack system components:
pRhlR-RhlL contains transcriptional units (TU's) of the QS system.
pRhlI is necessary for the production of the QS C4-HSL and RhlR encodes for the transcription factor that binds C4-HSL and the conditional promoter LasB.
pLDH conditionally expresses LDH.
pdCas9 is responsible for conditional expression of dCas9 in response to the QS system.
pMultiplex contains four constitutively expressed sgRNAs targeting multiple loci.
All vectors were designed using Benchling.
Each of the plasmids was to be constructed and optimized by running an individual DBTL cycle.
We chose to show our proof of concept in E. coli because it is both well studied and easy to engineer.
However, if our concept would prove to be successful the system could be transferred to different bacterial species naturally present in the vaginal microbiome, such as L. gasseri or L.Crispatus.
We chose to use the Rhl quorum sensing system from Pseudomonas aeruginosa, which is small, allowing for fast diffusion and functions in both gram positive and gram negative bacteria [1.].
We chose to isolate three components of the QS system that can function autonomously.
The molecule (C4-HSL produced by RhlI) and a transcription factor (RhlR) that activate the conditional promoter LasB [2.].
The quorum sensing system should be chromosomally integrated using pOSIP-CH.
Assembly of RhlR TU in pJUMP29-1A and RhlI TU in pJUMP29-1B.
The resulting vectors were successfully assembled into the pJUMP49-2B using a linker to substitute for the missing fragments (Level two plasmids usually contain four TU's).
However, transfer of the resulting plasmid
into
.
Thus no chromosomal integration was possible.
|
||||
| Promoter: BBa_J23106 |
RBS: BBa_K2680529 |
CDS: BBa_C0171 |
Terminator: Bba_J435371 |
|
| Promoter: BBa_J23106 |
RBS: BBa_K2680529 |
CDS: BBa_K4662044 |
Terminator: Bba_J435371 |
|
| Backbone: pJUMP49-2B(sfGFP) | ||||
Due to the difficulties we faced during chromosomal integration, we tested the functionality of the complete QS system by testing both individual components and the entire system through different co-transformations.
Recombinant production of C4-HSL by RhlI was tested using FIA-MS.
Figure 3. Spectrum of FIA-MS measurement:
The intensity is measured 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
Sensitivity of RhlR and LasB were tested by assembly of a plasmid containing RhlR and CFP (cyan fluorescent protein) under the expression of the QS dependent promoter LasB. was manually induced.
 |
||||
| Promoter: BBa_J23106 |
RBS: BBa_K2680529 |
CDS: BBa_C0171 |
Terminator: Bba_J435371 |
|
| Promoter: BBa_R0079 |
RBS: BBa_K2680529 |
CDS: BBa_K4662027 |
Terminator: Bba_J435371 |
|
| Backbone: pJUMP49-2B(sfGFP) | ||||
Figure 4.
Left: Raw measurement of fluorescence. The untransformed E. coli also has a fluorescent signal at 429 nm excitation.
Right: Values corrected for the fluorescent signal of the negative control.
After successfully testing the RhlR +CFP level 2 plasmid by manually induction and the positive RhlI level 1 plasmid expression and diffusion results, we co transformed both constructs into one bacterium. With the CFP output, we measured the fluorescent signal over time.
Figure 5.
Fluorescent signal at 429 nm excitation of the transformed but uninduced E.coli compared tothe transformed and induced E.coli.
Figure 5 confirms that the CFP signal increases as soon as the population density of the bacteria increases. Simultaneously with the production of the C4-HSL molecule.
Fluorescence was higher in induced plasmids confirming the functionality of the QS system.
The however the difference in fluorescence is relatively small compared to background noise.
That led to the conclusion that LasB showed , especially in the exponential phase of the bacteria when they are highly metabolically active.
To increase the sensitivity of the system a different Rhl dependent promoter could be used in the future [4.],[5.].
After complete characterization of our rhl-las quorum sensing hybrid system, we were able to confirm that RhlR is also able to induce the LasB promoter and the CFP moiety can be easily switched to the sequence of the protein of interest.
The endogenous LDH of E. coli we chose to express L-LDH from Steptococcus bovis.
It has high efficiency (kcat/Km value) and has previously been used for overproduction of Lactic acid in Escherichia coli [1.].
The sequence should be chromosomally integrated using pOSIP-TT.
L-LDH under the expression of LasB successful assembly into pJUMP29-1C. To compare the inducible LDH expression, we assembled a plasmid where the LDH is constitutively expressed pLDHconst, which was also transformed in the knockout strain.
To characterize the function of L-LDH under the expression of LasB we assembled a plasmid containing RhlR and LDH. The plasmid was induced by manually adding C4-HSL and the resulting lactate production was measured. To screen for minimal induction concentrations, a dilution series of the C4-HSL molecule was performed.
The constitutive expression of LDH was tested.
Figure 6. Measurement of L-lactate abundance in control strains; untransformed and knock out, as well as constitutive expression. Comparison to induced L-LDH via LasB promotor.
The constitutive promoter expressing LDH did not result in higher production of lactate compared to the control strain. As a next step, we would have exchanged the promoter for a stronger unit and repeated the experiment to have a better positive control.
As anticipated, the activation of the LasB promoter produced favorable outcomes, consistent with its prior characterization involving the expression of CFP.
Due to the fact that dCas9 shows [1.] effect [2.],[3.] when expressed at a high level we aimed for an dependent expression, which is activated by the LasB promoter.
LasB is much leakier than the Tet promoter used in many dCas9 systems.
To reduce the presence of dCas9 in the uninduced state we decided to add a degradation tag to increase the turnover rate of the protein.
To further optimize the level of dCas9 present in the cell we screened different RBS's we designed using the Salis lab RBS calculator [4.].
Additionally, we made a RBS library by randomly mutating four bases as described by Bikard et Al. [1.].
To select the optimal RBS we designed a blue white screen (Figure 7).
For this purpose, we constructed an additional plasmid containing RhlR and a sgRNA (pRhlR-LacZ) targeting the LacZ operon.
Expression of dCas9 with variable RBSs (pdCas9) in combination with the LacZ1-3 sgRNA should result in the silencing of the LacZ operon resulting in white colonies on X-Gal plates.
The desired amount of C4-HSL can be added to the system manually to induce the expression of dCas9.
Colonies that appear blue in the absence of C4-HSL (no silencing in uninduced state), and white in presence of C4-HSL (silencing in induced state) carry a plasmid with ideal RBS strength.
Figure 7. Schematic representation of blue white screen
See the Assembly Scheme here.
Vector Assembly
We successfully assembled different versions of
containing rationally designed RBSs and plasmids containing the RBS library.
|
||||
| Promoter: BBa_R0079 |
RBS: Variable- BBa_K4662000 BBa_K4662001 BBa_K4662002 BBa_K4662003 |
CDS: BBa_K4662061 |
Terminator: BBa_K4662004 |
|
| Backbone: pJUMP29-1C(sfGFP) | ||||
Blue white Screen
LacZ sg RNAs were tested by co transformation with pFD152 a plasmid containing dCas9 under the expression of a Tetracycline dependent promoter [4.].
Blue white screen was repeated several times using different batches of X-Gal plates and different levels of tetracycline as well as different methods for induction by tetracycline.
However, colonies always appeared blue.
Co-transformation of a plasmid containing LacZ sgRNA and RhlR ( ) with our QS dependent dCas9 plasmid also failed to produce white colonies.
 |
||||
| Promoter: BBa_J23106 |
RBS: BBa_K2680529 |
CDS: BBa_C0171 |
Terminator: Bba_J435371 |
|
| Promoter: BBa_J23102 |
RBS: BBa_K4662018 |
CDS: BBa_K4662031 |
Terminator: BBa_K4662007 |
|
| Backbone: pJUMP49-2B(sfGFP) | ||||
Figure 8. Unsuccessful induction of the LacZ sgRNA to block the beta-galactosidase
Blue white screen of LacZ sgRNAs could have failed due insufficient activation of the Tet promoter that regulates dCas9. To ensure sufficient activation, anhydrotetracycline which shows a stronger binding affinity and is not toxic to bacteria could be used to induce dCas9 expression. However it is also likely that the LacZ sgRNAs performed poorly. This is supported by the fact that co transformation of a plasmid containing LacZ sgRNA and RhlR with our QS dependent dCas9 plasmid also failed to produce white colonies. In a repeat experiment a different set of sgRNAs should be designed.
Optimal targets for flux redirecting were determined by dry lab modeling and verified using literature review [1.]-[13.]. The final targets were PflA, PflB and adhE. Additionally, we planned to knock out pta and ackA by chromosomally integrating at the corresponding site [14.]. The sgRNA were designed with the online tool chop chop. The three gRNA were selected to target the promoter or the beginning of the gene, and have a high predicted efficiency [15.].
| # | pyruvate formate-lyase 1-activating enzyme (pflA) | formate acetyltransferase 1 (pflB) | bifunctional aldehyde-alcohol dehydrogenase (adhE) |
|---|---|---|---|
| 1 | 76.91 | 70.06 | 54.97 |
| 2 | 68.36 | 64.83 | 70.24 |
| 3 | 67.99 | 69.98 | 69.67 |
Table 1. Predicted efficiency by chopchop
With PCR and our designed primers the gRNA should be adapted so that the promoter and an overhang for scarless assembly with the scaffold will be annealed.
Designed primers for amplification out the homology arms for AckA out of the bacterial genome.
| AckA upstream | |
|---|---|
| Forward | CGTCTCAggagCGTCTTTGAGTAATGCTGTCCC |
| Reverse | CGTCTCAggctGGAAGTACCTATAATTGATACGTGGC |
| AckA downstream | |
|---|---|
| Forward | CGTCTCAttcgTTTCACACCGCCAGCTCAGC |
| Reverse | CGTCTCAagcgCCTTCAACCAGAACGACTTCAGCG |
Figure 9. 1.5% agarose gel with 1kb ruler with the sgRNA after PCR
PCR was performed and the product was loaded on a gel, followed by gel purification. After sending it for sequencing, we realized that our assembled fragment is too short to perform the Gel or PCR purification with the spin column (only for fragments > 50bp). We tried the PCR again and performed an ethanol purification. We sent the samples again for sequencing and only one positive result came back.
Figure 10. Positive sequencing result after PCR and ethanol purification
Therefore the gRNA were ordered as a whole construct which already contains the promoter, scaffold and terminator. With the new version all level 1 sgRNA plasmids were successfully assembled and transformed.
Figure 11.
Left) Standard curve for D-lactate measurement.
Right) The lactate production transformed bacterium was measured in three conditions:
“uninduced” as a baseline how much lactate is produced normally and “induced” by either 0.5ug/ml and 5 ug/ml tetracycline.
We learn from our preliminary data that the sgRNA were able to block the enzymes, resulting in a higher lactic acid concentration when the dCas9 was induced with 5mg/ml tetracycline. Additionally we concluded that higher induction with tetracycline is not toxic for the cells. For future experiments, only the high induction concentration will be used.
We tested the sgRNA level 1 plasmids again.
Figure 12. The figures (a-i) show the corresponding lactate concentration at each timestep, in both conditions (uninduced and induced).
The predicted efficiency which was made at the beginning of the designing, had only a difference of around 10%. With our results, we are neither able to support nor deny the same efficiency order as the predictions.
After testing all the level 1 sgRNA constructs, we selected four sgRNA which showed the best effect for our level 2 plasmid. The following level 1 where chosen:
Based on the results obtained from the individual sgRNA screening we decided to assemble PflA1, PflB1, AdhE1 and AdhE2 into pMultiplex. For the guide RNAs of PflB and AdhE, there were nearly no differences measured in their activity. For PflA, the graphs show that the efficiency might be higher for PflA2 or PflA3, however we can see that the standard deviation is quite big compared to the PflA1 guide RNA. We focused on the most robust data, which also correlated with the prediction made by chopchop that PflA1 is the most efficient guide RNA. PflB1 was chosen because it has a highest predicted efficiency and its targeting sequencing in the promoter region. For AdhE we chose AdhE1 with a targeting area in the promoter and AdhE2 with the highest predicted efficiency.
(Level 2 sgRNA) was assembled with , , and .
|
||||
| Promoter: BBa_J23118 |
crRNA: BBa_K4662021 |
tracrRNA: BBa_C0171 |
Terminator: BBa_K4662010 |
|
| Promoter: BBa_J23118 |
crRNA: BBa_K4662022 |
tracrRNA: BBa_K4662031 |
Terminator: BBa_K4662010 |
|
| Promoter: BBa_J23119 |
crRNA: BBa_K4662012 |
tracrRNA: BBa_K4662031 |
Terminator: BBa_K4662008 |
|
| Promoter: BBa_J23100 |
crRNA: BBa_K4662014 |
tracrRNA: BBa_K4662031 |
Terminator: BBa_K4662009 |
|
| Backbone: pJUMP49-2B(sfGFP) | ||||
 |
||||
| Promoter: BBa_J23118 |
crRNA: BBa_K4662021 |
tracrRNA: BBa_K4662031 |
Terminator: BBa_K4662010 |
|
| Backbone: pJUMP29-1D(sfGFP) | ||||
 |
||||
| Promoter: BBa_J23118 |
crRNA: BBa_K4662022 |
tracrRNA: BBa_K4662031 |
Terminator: BBa_K4662010 |
|
| Backbone: pJUMP29-1D(sfGFP) | ||||
 |
||||
| Promoter: BBa_J23119 |
crRNA: BBa_K4662011 |
tracrRNA: BBa_K4662031 |
Terminator: BBa_K4662008 |
|
| Backbone: pJUMP29-1B(sfGFP) | ||||
 |
||||
| Promoter: BBa_J23100 |
crRNA: BBa_K4662015 |
tracrRNA: BBa_K4662031 |
Terminator: BBa_K4662009 |
|
| Backbone: pJUMP29-1C(sfGFP) | ||||
For the testing we furthermore increased the sample volume for the testing, and decreased the water content in the buffer to have an increased sample: buffer ratio. Since we switched the buffer conditions, we tested the level 1 plasmids as three biological replicates and two technical replicates.
Surprisingly the effect observed in Figure 14 was less prominent than we had expected. The standard deviations are very small, which supports that the knocking down of multiple enzymes simultaneously leads to a constant and efficient flux redirection towards lactate.
Figure 14. Assembly of PflA1, PflB1, AdhE1 and AdhE2 sgRNAs to knock down multiple genes at the same time.
The combination of the guide RNAs leads to more available pyruvate in the cell. This allows future users to redirect the flux of e.g.: glycolysis into a desired direction.
To characterize the functional sgRNAs in more detail, a mobility shift assay could be performed.
We managed to assemble and test all the components necessary for the function of our final product.
This includes:
Due to failed chromosomal integration, we were unable to test all of the components of our system in one bacterium.
However, chromosomal integration in E. coli is performed on a routine basis, and is likely to succeed in a repeat experiment.