Kill Switch
Alexa DeKorte and Christina Harbin
Why We Need a Kill Switch
Our team wanted to incorporate an added level of safety by adding a kill switch to the genome. It can be dangerous for modified bacteria to be released into the environment, so with the addition of the mazE/mazF toxin/antitoxin kill switch we can program the bacteria to be toxic to itself if it escapes its contained environment. In our designed plasmid, the toxin is always expressed, as well as the expression of the activator molecule HSL. This molecule induces the expression of the antitoxin, so when there is a high enough concentration of cells expressing HSL, the antitoxin will be expressed. This antitoxin will bind to the toxin to prevent cell death and keep the cell alive. Our bacteria are programmed to self-destruct at a lower concentration. Such concentration would occur if these bacteria were to spill into the environment. If a spill were to occur and the concentration of cells was to decrease, there would not be enough HSL to activate the mazE anti-toxin. Therefore, the toxin mazF would kill the cell.
Our Approach to a Kill Switch
We designed a kill switch for our bacteria in order to ensure that an engineered bacteria could not survive outside of its contained environment. We did this by using the mazE/maxF toxin and antitoxin genes. As shown in the figure below, when LuxI and LuxR are expressed, HSL is produced. This AHL serves as an activator for the pLux promoter that leads to the expression of mazE, an antitoxin to mazF. When LuxI and LuxR are expressed, mazF is also expressed. This is a toxin that causes cell death. When there is a high enough concentration of AHL, the anti toxin is then expressed and it binds to mazF preventing cell death. If there were ever a spill, the concentration of cells and therefore of AHL would be too low to activate the pLux promoter and the mazF would then be able to initiate cell death.
Genetic Circuit:
Design and Problem Solving:
We found each of the pieces in the circuit individually on the parts registry and copied the nucleotide sequences of the circuit together. We split the circuit into pieces with approximately 700 bases each, with a small overlap of nucleotides in order to get the DNA to assemble properly.
We ran into a problem when running the sequence through IDT-DNA. There was an issue with the ribosome binding site (BBa_B0030) being in multiple locations. This issue was combated by designing the pieces of DNA so that they were all approximately 700 bases long, but only contained 1 ribosome binding site per piece of DNA.
Experiments
Parts and Assembly:
The plasmid backbone was designed with bases complementary to those at the ends of the circuit in order to guide proper insertion of the circuit into the plasmid. The designed plasmid went through the PCR protocol for amplification. Pictured below is a gel electrophoresis showing bright bands at the location representing the proper number of base pairs contained in our plasmid. These bands are found between 2kbp and 3kbp.
After ordering the pieces of DNA, we used the HiFi assembly method to amplify and assemble the DNA in the circuit in order to complete a transformation. We attempted the transformation in NEB 5-alpha competent cells. The cells were plated on agar plates so that the bacteria they were transformed into could grow and be analyzed. In order to maintain the optimal conditions for the growth and development of these bacteria, we added a concentration of 10uM of HSL to both the plates and the competent cells. This value was determined based on the data for LuxI/LuxR and pLux promoter parts on the iGEM parts registry (part:BBa_C0062). This was in order to be sure activation of the pLux promoter occurred so that enough of the antitoxin could be produced to prevent cell death.
Backup reactions using the same conditions were set up, and contained other diagnostic/ control copies of the kill switch. This was so that we could more accurately assess the output of the circuit.
The transformed cells, including its backup replicates, were plated on the HSL enhanced agar plates and incubated. Colonies of bacteria grew and we conducted colony PCR to use for gel electrophoresis in order to quantize correct synthesis of the plasmid. We also did a streak plate with (4) colonies from the control plate and (4) colonies from the mazE/mazF circuit. This was to have access to the colony to send for sequencing should the gel indicate an accurate number of bases.
For the backup reactions, colonies were observed and we also ran a colony PCR for these reactions by choosing 23 colonies to run a PCR on, as well as a streak test. There were 4 different designed plasmids that were transformed into bacteria: The actual mazE/mazF kill switch (the plasmid containing RFP in the place of the toxin, the plasmid containing GFP in the place of the toxin, and the plasmid replacing both the toxin and the antitoxin with RFP and GFP respectively). We chose 23 random colonies, 6 from each type and only 5 for the plasmid replacing both the toxin and the antitoxin with RFP and GFP respectively, and ran a colony PCR that was analyzed. We created a master mix of 125 uL of Taq, 50 uL of VR, 50 uL of VF2, and 25 uL of ddH2O. We then added 10 uL into a PCR tube and added the colony. We also replated each colony using a streak test. We then put them into the PCR machine at 59 degrees, at 30 cycles, with an annealing time of 160 seconds.
The colonies that were expected to have proper synthesis were grown in a liquid culture so that they could be used to characterize the plasmid if the sequencing confirmed synthesis. We ran a gel electrophoresis for all of the samples and sent them to MSU for sequencing to confirm synthesis of the plasmid.
Colony plates from original reactions and colonies chosen for colony PCR (1-8)
Colony plates of backup reactions chosen for colony PCR (as seen above) (1-23)
Implementation
Testing
After colony PCR, a gel electrophoresis was run for each colony and above are images of the gels with numbers indicating bands that correspond to the appropriate number of base pairs for our given circuit. These samples included α1 and α5, designed to be our intended kill switch, 6 and 8, which were both the control switch containing RFP in the place of the toxin, and 13, which was a control copy of the kill switch replacing the toxin/antitoxin with RFP/GFP respectively.
Gel electrophoresis for reactions 6 and 8, suspected to contain control kill switch with RFP replacing mazF.
Gel electrophoresis for colonies a1 and a5, which are suspected to contain the mazE/mazF kill switch.
Gel suspected to contain colony 13 which is suspected to have GFP/RFP mazE/mazF diagnostic control switch.
In addition to the gel indicating synthesis, the expression of GFP/ RFP was observed visually in the control kill switches (see right). Colony 6 and 8 had visible red coloring, while colony 13 was a distinct orange color often associated with expression of GFP and RFP simultaneously. This also indicates that some synthesis was promising. The plasmids suspected to be properly synthesized were purified using the QuickLyse Miniprep Kit and then sent to MSU for sequencing.
Based on the analysis of the sequencing results through NCBI BlastX, we had (2) working copies of the designed kill switch, 2 working copies of the control kill switch containing RFP in the place of the toxin. Samples 9 and 10, colony 13, did not have good sequencing results, however, the color of the colony as well as the colony PCR indicated that there was also 1 copy of the control kill switch replacing the toxin and antitoxin with RFP and GFP respectively.
| Sample | Colony | Primer | Expected | NCBI BlastX results | Alignment Results |
|---|---|---|---|---|---|
| 1 | Alpha-1 | VF2 | MazF kill switch | mazF, mazE | good |
| 2 | Alpha-1 | VR | MazF kill switch | mazE | good |
| 3 | Alpha-5 | VF2 | MazF kill switch | none | bad |
| 4 | Alpha-5 | VR | MazF kill switch | mazF, mazE | good |
| 5 | 6 | VF2 | Control kill switch (RFP) | LuxI | good |
| 6 | 6 | VR | Control kill Switch (RFP) | LuxI | good |
| 7 | 8 | VF2 | Control Kill Switch (RFP) | MazE | good |
| 8 | 8 | VF2 | Control Kill Switch (RFP) | LuxI | good |
| 9 | 13 | VR | Control Kill Switch (RFP) | none | bad |
| 10 | 13 | VF2 | Control Kill Switch (RFP) | none | bad |
Proof of Concept:
The next step was confirmation of synthesis through other means and testing the function of each switch under different conditions. We prepared RNA from our competent cells using a RNA Mini Kit to become cDNA for qPCR to further confirm synthesis. We did this by running a qPCR with primers targeting specific genes in the sequence to determine how many PCR cycles it takes to reach a cDNA threshold. The lower the number of cycles means a greater amount of RNA/ cDNA expression in our plasmids. This is a way of quantitating the levels of expression for each specific part of the sequence. This further confirms proper synthesis, as each gene should have similar levels of expression to those on the same circuit. It also tells us how the levels of the toxin/ antitoxin/ control compare to each other, which is important data about the function of the kill switch. We ran 16 reactions and 5 controls, one for each of the 5 working parts, using RNA rather than cDNA to test for contamination.
Controls:
| Bacteria | cDNA primers (F/R) Tested |
|---|---|
| α1- mazE/mazF kill switch | mazF |
| α5- maxE/maxF kill switch | LuxR |
| 6- RFP control | mazE |
| 8- RFP control | RFP |
| 13- GFP/RFP Control | GFP |
qPCR Reactions run:
| 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| cDNA | α1 | α1 | α1 | α1 | α5 | α5 | α5 | 6 | 6 | 6 | 8 | 8 | 8 | 13 | 13 | 13 |
| Primer Type | I | R | F | E | R | rf | E | I | rf | E | I | rf | E | R | gf | rf |
F= mazF, E=mazE, I=LuxI, R=LuxR, rf= RFP, gf=GFP
qPCR expression results:
The graph shows the CT values for Lux R (gray), Lux I (green), mazF (blue), and the control (red). Lux R, Lux I, and mazF are all on the same RNA strand so their CT values should be the same. However, the results show higher CT value for mazF, indicating less RNA. From this we could infer that the RNA is prematurely terminating, resulting in less mazF RNA. LuxI and LuxR have very similar CT values indicating that LuxI has not problem going into LuxR. The control has the highest CT value meaning it has a lower starting amount of nucleic acid. This is because we didn’t convert it to DNA. What we see on the graph is what is left over from purification.
This graph shows us 1 peak for Lux R (gray), Lux I (maroon), and mazF (blue) meaning only one DNA molecule present. This was used to confirm that the graph above is usable. Multiple peaks for a single reaction would mean we would have to disregard the CT values.
As an analysis of the suspected termination of the circuit, the sequence was run through ARNold to determine if there are any predicted transcription termination sequences hidden. This analysis did not indicate/ detect any of these sequences, so it’s lack of expression could have another explanation.
To test the efficacy and confirm the action of the kill switch, varying levels of HSL were used to prepare agar plates, which colonies were transplanted onto. The plates were either 10uL or a control plate. The top two colonies on the picture to the right were suspected to have successfully integrated the kill switch. This can be seen as there is slightly less growth in the first two columns of the first two rows on the control plate. Also, the colonies on the control plate took longer to grow. This difference is not great indicating that our kill switch works but isn’t very effective.
Results
Overall, we did in fact synthesize the kill switch we designed, however, it may not work effectively yet. After running the qPCR targeting for specific genes, the analysis indicates that mazF may be degrading at its current location in the circuit. The expression of LuxR and LuxI are at similar levels, however there is decreased expression of the toxin. To be an effective kill switch, expression of mazF would need to be increased, while currently it is suspected to be terminating early/ degrading at the end of the gene leading to lower expression. These results are consistent with the testing of function at different concentrations of HSL and different concentrations of cells in the spot. From the spot testing, it is shown that there is still growth at lower cell concentrations even when no HSL is present. While growth visually does look to be slightly lower for that of alpha1/alpha5 when no HSL is present and at the lower concentration of cells, we do still see growth. The spot testing revealed that even at no concentration of HSL, the colonies can still grow which is not our intention. We would not like to see growth when no HSL is present because growth means that there is not proper function of the toxin in the circuit, if it were working as intended, there would be no cell growth at low cell and/or HSL concentrations. The kill switch did in fact synthesize, but is currently in need of fine tuning.
Growth Curve for Strains
We wanted to quantitate our observations from the plates, and so we inoculated 3mL of LB with chloramphenicol, but without any AHL molecules, with different amounts of the starting cultures.
Starting from an overnight culture, we pelleted the cells by centrifugation, washed them to remove any residual AHL molecules, and resuspended them at an OD600 of 0.1 (approximately 1x109 CFU/mL). Then, we used this to inoculate our fresh LB at two different levels – a relatively high density (OD600 of 1x10-5 or about 1x104 cells) or a lower density (OD600 of 1x10-8, or about 10 cells). We tracked their growth by optical density over the next day (If an OD of zero was read, we assumed it was 0.0005, which is below the limit of detection for our spectrophotometer).
| Time | 0 | 12 | 24 | 28 |
|---|---|---|---|---|
| KS-1 High | 1.00E-05 | 0.1740 | 3.7250 | 4.8350 |
| KS-1 Low | 1.00E-08 | 0.0005 | 1.9050 | 3.5050 |
| KS-5 High | 1.00E-05 | 0.0600 | 4.0200 | 4.2600 |
| KS-5 Low | 1.00E-08 | 0.0005 | 1.9100 | 4.2500 |
| RFP-6 High | 1.00E-05 | 0.8480 | 4.2400 | 4.2450 |
| RFP-6 Low | 1.00E-08 | 0.0005 | 3.5450 | 3.6350 |
| GFP-13 High | 1.00E-05 | 0.1140 | 4.8200 | 5.1000 |
| GFP-13 Low | 1.00E-08 | 0.0005 | 2.4450 | 4.5250 |
KS represents two clones of our kill-switch, K4166000. RFP-6 is a clone of one of our controls (K4166001) in which MazF is replaced with RFP. GFP-13 is another control clone (K416014) in which, in addition to the MazF/RFP substitution, MazE is replaced with GFP.
From this, we can see that the kill-switch stunts, but does not prevent, bacterial growth.
Next Steps
Our next steps include improving our kill switch to be more effective. MazF is prematurely terminating and there are a few ways we can try to fix this. We could rearrange the circuit so that maz F is before LuxI and LuxR in the circuit. We know there is no problem with LuxI going into LuxR so a new rearrangement of the circuit could be mazF to LuxI to LuxR. This would still lead to expression of HSL, but the expression of the toxin would be increased if it was not located at the end of the circuit/has less chance to be degraded. We also could increase the strength of the promoter located before the LuxI, LuxR, and mazF. We would do this by researching the efficacy and function of previously characterized promoters and choosing one that is more suited for the behaviors we are attempting to achieve from our circuit. Lastly, we could put a more active ribosome binding site before mazF and/or a less active binding site before mazE. This would lead to increased/stronger expression of mazF or less expression/strength of the anti toxin.