What is a Toxin-Antitoxin System?
A bacterial toxin-antitoxin (TA) system is based on a genetic module that includes a gene pair, consisting of a stable toxin gene and an unstable antitoxin gene. Such systems exist either on extrachromosomal elements or on bacterial chromosomes. In general, toxin-antitoxin systems can be classified into six common types .
We utilize a naturally present system in E. coli, the mazF/mazE toxin-antitoxin (TA) pair. We specifically chose this system because it is known as the most studied TA system and therefore we had a lot of research to base our project on. Toxin-antitoxin systems occur naturally in bacteria and can be necessary for plasmid retention as well as they play a crucial role in plasmid retention . The toxin is often more stable than the antitoxin and lethal to the cell. Only the presence of the antitoxin neutralizes the toxin, leading to the survival of the cell. In the case of plasmid loss, both the toxin and antitoxin are not produced anymore. The antitoxin is degraded quickly, leaving the cell with only toxin, which leads to cell death. MazE/mazF is a natively occuring system and part of a stress reaction that pauses cellular activity to conserve energy and resources.
MazF serves as a toxin, functioning as an endoribonuclease, capable of cleaving both single and double stranded RNA. As soon as the concentration of the toxin is higher than what the antitoxin can neutralize, the toxin leads to growth inhibition. The antitoxin protein mazE neutralizes the toxin by forming a protein-protein complex in the form of a heterohexamer. It consists of two mazE monomers, which are bounded with four mazF monomers. The naturally occurring toxin-antitoxin system is also present in the strain we are using. Because of this, a constitutive level of toxin-antitoxin expression is given. The endogenous toxin-antitoxin complex regulates its own gene expression , thus we decided to use a different promoter than the native one to prevent interference by the toxin-antitoxin complex. To gain insights on the functionality of the toxin-antitoxin system we performed several experiments.
In order to protect our system from high toxin concentrations it is designed to constitutively produce an antitoxin (mazE) to neutralize the toxin (mazF) during the initial stage of B12 production. Once produced, B12 inhibits the toxin transcription and translation via binding to the negative regulated riboswitch, leading to cell survival and increased production of B12. In the absence of sufficient B12 production, toxin levels rise due to an inactive riboswitch structure, leading to cell death.
Short overview of how our general workflow was and what we successfully achieved during our wetlab. The main measurement principles from which we have drawn our toxin results are listed below.
Test Toxin Production
Determining if cells expressing mazF (toxin) undergo cell death and in what time range this takes place.
To test the toxin, we used pGGAselect as a backbone, containing the TetR promoter, a riboswitch, the mazF toxin fused to a fluorescent protein (mTurquoise) with a N-terminal FLAG-Tag for detection (piG_02). In parallel, the experiment was performed with a plasmid (piG_03) containing mazF without the fluorescent protein. As a negative control we used the empty pGGAselect.
Since the riboswitch is negatively regulated by AdoCbl, which is present in small amounts in LB medium, we use M9 medium that does not contain AdoCbl. Since there is no AdoCbl that can bind to the riboswitch and regulate toxin expression via a conformational change, it is possible to test toxin expression independently of the riboswitch. The plasmids were transformed into E. coli MG1655. For the pre-cultures we then transferred the bacteria into flasks with 300 mL M9 medium and 34 mg/mL chloramphenicol. The cells grew at 37°C for about two days until an OD600 = 0.5/0.6 was reached. After dividing each culture into 6 flasks containing 50 mL of culture, they were induced with 0, 10, 50, 100 and 500 ng/mL DOX. For the next steps, we measured cell growth (OD600) and fluorescence (Ex: 434 nm; Em: 474 nm) of each probe after 0, 2, 4, 6, 12, 24, 36 and 48 hours with the Plate Reader. Additionally, 2 mL of culture were taken at each time point, 1 mL for Western blot analysis and 1 mL to determine forming unit (CFU). The samples for Western Blot were centrifuged at 12000 rpm for 2 minutes, the supernatant was discarded and then frozen at -80°C. For the CFU measurements, 1 mL of each culture was centrifuged at 8000 rpm for 3 min, cell pellets washed with 1 mL DPBS and resuspended in 1 mL LB medium. We diluted all samples to OD600 = 0.1 and then performed a serial dilution. The final dilutions (10-4 and 10-5) were plated on LB plates with chloramphenicol.
After an incubation time of 12-16 hours the survival rate of the cells was measured by counting the colonies. CFUs were calculated for each sample. In addition we did fluorescent measurements of mazF-mTurquoise expression to test whether and how many cells express the fluorescent protein and thus also the toxin mazF. Therefore, we took samples at 0 and 3 hours after induction. For the microscopy pictures we used the Zeiss Axio Observer7 with Colibri LED to capture bright field and fluorescent images with a FITC filter (474 nm) Images were processed and analyzed with Fiji Imaging processing software.
No growth inhibition was observed for cells containing piG_02, where mTurquoise is fused to mazF, induced with 0, 10, 50, 100 and 500 ng/mL DOX. Similar OD600 values at each time point indicate a similar growth behavior (Figure 3). Cells containing piG_03 in the non-induced and induced state of DOX (50, 100,500 ng/mL) showed a slightly decreasing growth behavior only after 48 hours (Figure 4).
An increase in fluorescent signal was observed for piG_02 (toxin with mTurquoise) can be detected which suggests toxin expression (Figure 5). As control, piG_03 cells (toxin without mTurquoise) in Figure 6 showed no difference in fluorescent signal over time in the non-induced and induced state (10, 50, 100, 500 ng/mL).
We used fluorescent microscopy to visualize the expression of toxin fused to mTurquoise. Samples were taken from the cultures containing piG_02 in E. coli (MG1655) induced with 100 ng/mL DOX, 2 hours and 4 hours after induction. We observed that most of the cells did not produce visible amounts of toxin, mazF-mTurquoise (Figure 7). While after 2 hours almost no fluorescent cells could be observed, we were able to see some mTurquoise positive cells 4 hours after induction. However, the majority of cells did not express mazF-mTurquoise.
To detect mazF toxin on a western blot, we fused a FLAG-tag at the N-term. Expression of mazF-mTurquoise, piG_02, fusion would correspond to a band with a size of 40 kDa, whereas expression of the toxin alone, piG_03, would correspond to a band of 13 kDa. In Figure 8A and 8B we show the western blot results for the samples taken from the experiments of Figure 3 and 4.
Figure 8A shows the samples for piG_02 after 12h (lane 2-6). Here, we observe a slight band at 40 kDa for the samples induced with 50-500 ng/mL. However, the quality of this part of the gel was not optimal, as we cut the membrane in the area where we would expect the band. Therefore, we reran the samples on another gel and included more timepoints (Figure 8B). Replicate 2 presented in Figure 8B shows expression of mazF-mTurquoise fusion with a size of 40 kDa for piG_02 cells, induced with 10, 50, 100 and 500 ng/mL DOX after 8 and 12 hours and no protein expression in the non-induced state after 6 and 12 hours.
The western blot with the samples of piG_03 (Figure 8A, lane 7-11) shows a band at 40 kDa after 12 hours. However, we would have expected a band at 13 kDa. This was most probably due to pipetting the wrong samples on the SDS gel. Therefore, we decided to rerun the samples for piG_03 on another gel (Figure 9).
We observe a strong band in the samples taken after 12 hours of about 13 kDa, corresponding to the size of MazF. There was no mazF observed without induction with DOX at either 0 hour or 12 hours after induction. Unfortunately, we also did not detect any protein when induced with 100 ng/mL DOX. This was most probably due to pipetting error. We also observe for all samples a band around 35 kDa. This seems to be an unspecific binding to a cellular protein as it is present also in the uninduced samples.
In conclusion, we did not observe significant cell death or growth inhibition when measuring OD600 (Figure 3 and 4) even for higher DOX concentrations. Unfortunately, we were not able to evaluate the CFUs from this experiment because there was a bacterial lawn on the plates due to the high growth rate. Those experiments were repeated three times without any growth inhibition or cell death observed. During this time period, the question arose to what extent we prevent the cells from mutating the toxin, or the associated promoter. Therefore, we screened by sequencing several generations of bacteria containing piG_02, which had been induced to express the toxin.
According to the sequencing results of the toxin we had no mutations, which could have affected the function of the toxin. Apparently, the concerns of toxin mutation do not seem to have any impact on our system in the period we are looking at. However, when we measured fluorescence we saw an increase in fluorescence with higher DOX concentrations (Figure 5). This would suggest that the lack of cell death is not due to missing expression. This was also confirmed by western blot (Figure 8 and 9) where we detected bands for both toxin-mTurquoise fusion as well as the toxin alone.
However, the increase in the intensity of the bands with higher DOX concentrations was only mild. We hypothesized that this could be the result of a subculture in our glycerol stock that either does not produce toxin or has found a way to inhibit Toxin expression.
Fluorescence Microscopy of mazF-mTurquoise
We wanted to investigate whether the lack of fluorescent cells observed using microscopy was due to an evolving subpopulation capable of shutting down toxin-mTurquoise production, or if there was contamination of our glycerol stock. Therefore, we did fluorescence microscopy of single colonies from our glycerol stock.
Glycerol stock of piG_02 was plated on agar plates and left overnight in the incubator at 37°C. The following day single colonies were picked and grown according to the last experiments. The cultures were induced with 100 ng/mL DOX induced. Samples for microscopy were taken at time 0 and 3 hours after induction. For the microscopy pictures we used the Zeiss Axio Observer7 with Colibri LED to capture bright field and fluorescent images with a FITC filter (474 nm)
Figure 10 shows microscopy pictures of 2 different colonies. We observed that in contrast to the previous experiment almost every cell expressed mTurquoise. This, unfortunately, indicates contamination of our glycerol stock. Moreover, we observed expression of mTurquoise in samples at time 0, indicating a leaky promoter.
Upon microscopy of single colonies, we could conclude that there is a contamination of the glycerol stock (Figure 10). This very likely also had had an effect on our previous toxin experiments. In addition, we can conclude from the microscopy images that the promoter is leaky. The next step would be to prepare a new glycerol stock of piG_02 without contamination and repeat the growth assay.
However, as seen in Figure 10, the cells expressing high levels of the toxin-mTurquoise fusion 3 hours after induction still seem to survive. Moreover, the results of piG_03 (Figure 4) show only a slight tendency of cell death and only at late time points. This points to the possibility that the contamination was not the sole reason that the cells were not affected by toxin. Therefore, we designed a plasmid (piG_23) without the riboswitch upstream of the toxin mazF and the RBS from the TetR promoter (see plasmid page).
Production of Toxin varies between different Constructs
The aim of this experiment was to investigate the toxic effect of MazF expressed from different constructs. We tested the following constructs: piG_02, piG_03, piG_23 and the empty backbone pGGAselect. This experiment was carried out with a newly prepared glycerol stock of piG_02, as we were able to detect contamination in the old glycerol stock on the basis of microscopy images.
The experimental procedure remained the same as in the previous experiment. Due to the different amount of samples needed from each condition, we cultivated the cultures in different volumes. For pGGAselect cells we used a volume of 110 mL (two induction states: 0, 100 ng/mL), a volume of 160 mL each for piG_02 and piG_03 cells (0, 50, 100 ng/mL) and a volume of 300 mL for piG_23 cells (0, 10, 50, 100, 500 ng/mL). When OD600 = 0.5 was reached we splitted the cultures into several 50 mL flasks. In addition, a washing step was necessary for the CFU samples in order to spread the bacterial cultures on LB plates.
In contrast to the previous experiments, growth inhibition was observed for piG_23, when induced with 50, 100, or 500 nM/ml DOX (Figure 12). In Figures 13 and 14 no significant difference in growth could be observed for both plasmids that contain the riboswitch downstream of mazF (piG_02 and piG_03). However, a slightly lower OD600 of piG_03 induced with 100 ng/mL DOX was observed in comparison to the non-induced condition. The negative control of empty pGGAselect cells showed no growth inhibition, as expected (Figure 15).
However, as the experiment was only repeated once and because the differences in OD are small it is hard to determine if there is an actual effect of the toxin in this case. Nevertheless, it can be seen that the bacteria carrying piG_23 grew to a lower OD600 when induced with DOX in comparison to piG_02 and piG_03.
The CFU/mL measurements of each plasmid over time is shown in Figure 16. Overall, a decrease in CFUs can be observed over the measured period for piG_23, especially with an induction concentration higher than 50 ng/mL (Figure 16A). We did not observe any colony for piG_23 plasmid, following induction with 50, 100 and 500 ng/mL DOX. This confirms the observed stagnation in growth curves and could indicate cell death. In general, piG_23 cells non-induced and with 10 ng/mL DOX shows no differences CFU over time. Similarly, the negative control with the empty backbone pGGAselect (non-induced and induced) shows no signs of significant cell death (Figure 16).
We observed a bacterial lawn on the plates of piG_02 with 0, 50, 100 ng/mL DOX and piG_03 with 100 ng/mL DOX at time point 12 hours, which were not countable (marked with *). Therefore, these data could not be included. Overall, the results for piG_02 and piG_03 indicate that even induction with high DOX concentrations (50 ng/mL and 100 ng/mL) did not affect the survival of the cells. This is in line with what we have already seen in the growth curves (Figures 13 and 14).
We further verified the expression of mazF with western blotting (Figure 17). Two different buffers were used for the Western blot, Laemmli and non-degrading buffers. To determine whether the toxin is also present as a dimer in the cell, we used a non-degrading buffer. This allows native protein folding to be preserved to some extent.
In Figure 20A we see a strong band for the non-degrading samples containing piG_23 for all time points. For expression of the toxin alone we expect a band at 13 kDa, which fits with our observations. Moreover, we detected an increasing amount of mazF (13 kDa) with time of induced (100 ng/mL) piG_23 (Figure 17A). Interestingly, another band at around 26 kDa can be seen, possibly representing mazF dimer as samples were treated with a non-degrading buffer. Unfortunately, we did not detect any mazF with the same samples of piG_23 treated with Laemmli buffer (Figure 17B). However, a faint band could be observed for the piG_23 12 hours after induction with 50 ng/mL.
Figure 17C and D show the results for piG_02 and piG_03. Similarly to the results for the Laemmli buffer for piG_23 we see no or only faint bands. This was in contrast to the first western blots we made (Figure 9), however, we could not find an explanation for this.
For piG_02 the protein size is about 40 kDa due to the fusion to mTurquoise. We did not detect any protein for piG_02 in Laemmli and non-degrading buffer when induced with 100 ng/mL DOX after 0, 1, 4, 6 and 8 hours (Figure 17C). However, a band was observed after 12 hours after induction in Laemmli and non-degrading buffer (Figure 17C). The bands are not running at exactly the same height for Laemmli and non-degrading buffers but we expect the differences to be due to the folding of protein in the non-degrading buffer.
For piG_03, protein was detectable after 4 and 12 hours in Laemmli buffer as well as after 4, 6, 8 and 12 hours in non-degrading buffer (Figure 17D). piG_03 shows, according to the growth curve results (Figure 14), a weaker mazF expression compared to piG_23 which correlates with the weak bands at a length of 13 and 26 kDa.
The results suggest that the toxin plasmids with riboswitch, piG_02 and piG_03, have a significantly less toxic effect than expected. However, by removing the riboswitch (piG_23), we could show that overexpression of mazF leads to growth inhibition or cell death.
To investigate whether only the RBS is the cause of the low expression mazF, we designed another plasmid (piG_23b). We inserted the RBS of the riboswitch into piG_23. Moreover we planned to insert the strong RBS from piG_23 between riboswitch and mazF in piG_03, unfortunately we were not able to clone it. Based on literature we assumed that the riboswitch RBS is relatively weak . This could mean that even when the expression of the toxin is not repressed, still too little is produced. This could be one explanation why we don’t observe cell death. Indeed the piG_23 has a strong RBS. Next we made a short experiment with piG_23 cultured in LB. By culturing in LB the cultures grow faster and to higher ODs. Therefore, we would be able to confirm the findings of the toxic piG_23 in a fast and simple setup.
Toxin Production without Riboswitch leads to Cell Death
In this experiment, we tested the toxicity of the toxin mazF without the presence of the riboswitch in LB medium.
In the previous experiments, we tested the toxin in M9 medium instead of LB medium because LB contains AdoCbl. Since this plasmid contains only the toxin without the riboswitch, which is negatively regulated by AdoCbl and thus regulates toxin expression, it was possible to test the experiment in LB medium. In addition to piG_23, the empty backbone pGGAselect was used as control. We incubated the cultures until an OD600 = 0.5/0.6 was reached. After dividing each culture into 5 flasks containing 50 mL of culture, they were induced with 0, 10, 50, 100, 500 ng/mL DOX. For the next steps, we measured OD600 of each probe after 0, 1, 4, 6, 8, 12, 24, 48 hours. Additionally, 2 mL of culture were taken at each time point, 1 mL for western blot samples and 1 mL for CFU measurement. For western blot samples, the samples were centrifuged at 12000 rpm for 2 minutes, the supernatant was discarded and then frozen at -80°C. For the CFU samples, we first diluted our sample to OD600 = 0.1 and then diluted them to a concentration of 10-4 and 10-5. The final dilutions were plated on LB plates with chloramphenicol. After an incubation time of 12-16 hours the survival rate of the cells was measured by counting the CFUs.
This experiment further demonstrated the toxicity of our toxin. A significant growth inhibition was observed for piG_23 cells induced with 50, 100 and 500 ng/ml DOX (Figure 19). Constant OD600 values of around 1.0 indicate a stagnation in growth. Control cells containing pGGAselect showed a steady increase in growth over time in the non-induced and induced state (100 ng/mL). Cells containing piG_23 in the non-induced and induced state of low DOX (10 ng/mL) showed a similar growth behavior as the control pGGAselect, reaching an OD600 of about 8.5 (Figure 19).
Next we evaluated this observation by counting the CFU/mL (Figure 20). The CFUs showed a similar trend to the growth curve and showed a low to no colony count when induced with 50 ng/mL or higher. This clearly shows the toxicity of the toxin when induced with DOX concentrations of 50 ng/mL and higher. In general, the control with the empty backbone pGGAselect (non-induced and induced) shows similar trends in terms of CFU over time as piG_23 non-induced and with 10 ng/mL induction.
Due to overgrowth, some data could not be included, as the measurements of piG_23 non-induced at time point 8 and 24 hours as well as piG_23 with an inducer concentration of 10 ng/mL at time point 8 and 24 hours. This is because of the bacterial lawn on the plates due to a high growth rate which was not countable.
Figure 21 shows the western blot results of the samples at time point 12 hours for the different inducer concentrations. MazF has a molecular weight of 13 kDa. The results are similar for the samples prepared in non-degrading and Laemmli Buffer. At an inducer concentration of 50 ng/mL or higher MazF is expressed whereas only weak expression at lower inducer concentration (10 ng/mL) is observed. By using a non-degrading buffer it is possible to determine whether the toxin is also present as a dimer in the cell as it allows native protein folding to be preserved to some extent. We observe weaker bands at a length of 26 kDa which would correspond to the mazF dimer. We detect no protein expression in pGGAselect cells which are marked in red.
By exchanging the riboswitch with only a normal RBS, we could show that induced overexpression of mazF leads to growth inhibition and cell death. The result indicates that in piG_02/piG_03, where the riboswitch is downstream of mazF, including its own internal RBS, might not efficiently allow translation.
To prove that the toxic effect really depends only on the RBS and that the riboswitch sequence itself has no effect on the toxicity of the toxin, we cloned and tested piG_23 but with the weak Riboswitch RBS. In this case, the strong RBS in piG_23 will be compared with the weak Riboswitch RBS.
Comparison of two RBS and how they affect Toxicity
The aim of this experiment is to test whether the riboswitch RBS is ultimately the cause of the weak toxin expression and consequently the weak toxin effect. For this purpose, we designed an experiment with piG_23, which contains the strong RBS, and compared it with a re-cloned version of piG_23 (piG_23b). The only difference between these two plasmids is the RBS. In plasmid piG_23b the RBS originates from the Riboswitch itself. In this way, we can exclude that the riboswitch has a negative effect on toxin expression and therefore interacts with the toxin. To support our findings we want to compare the toxic effect on the cells of both plasmids by determining the cell length.
As in the experimental setups before, we cultivated the bacteria in M9 medium and let them grow in pre-cultures until they reached an OD600 = 0.5/0.6 before induction with 100 ng/mL DOX. We took microscopy images to test whether and how many cells show fluctuations regarding cell length. Therefore, we took samples of cells containing piG_23b each 0 hours and 3 hours.
The piG_23b containing cells, as well as control cells containing pGGAselect, showed a steady increase in growth over time in the non-induced and induced state with 100 ng/mL DOX (Figure 23). This result indicates no difference in growth, while significant growth inhibition can be observed for piG_23 cells induced with 100 ng/ml DOX, as we have seen in previous experiments (Figure 19). Constant OD600 values of around 0.8 indicate a stagnation in growth. Unexpectedly, we observed that piG_23 non-induced also showed signs of growth stagnation. Therefore, we decided to repeat the experiment.
Replicate 2 presented in Figure 24 showed a similar growth behavior as replicate 1. We observed that the bacteria carrying piG_23b reached a higher OD although being induced with the same DOX concentration as the bacteria containing piG_23. In contrast to the last replicate, the non-induced piG_23 grew as expected, reaching similar ODs as the non-induced pGGAselect (Figure 24). However, for this repetition, we observed an unexpected growth of induced pGGAselect. This could, however, not be explained and it was the first time we observed this.
Unfortunately, we were not able to evaluate the CFUs of replicate 1 because there was a bacterial lawn on the plates most probably due to a high growth rate and pipette mistakes. The CFU/mL measurements of replicate 2 of each plasmid over time is shown in Figure 25. There is no difference in CFU for both, piG_23b and pGGAselect, whether non-induced or induced with DOX (Figure 25B and C). Moreover, the lowest CFU rate for piG_23b and pGGAselect cells was determined at time point 0, 4 and unfortunately 24 hours after induction. This was most probably due to pipetting error or stressed cells (Figure 25B and C). Again, a decrease in CFUs can be observed over the time measured for piG_23 with an induction concentration of 100 ng/mL DOX. This confirms the observed stagnation in growth and could indicate cell death. (Figure 25A , 23 and 24).
After another round of researching we found a publication showing that mazF-mediated cleavage of mazF mRNA leads to increased temporal variability in length of cells in isogenic populations of E. coli , while overexpressing mazF. By taking microscopy images of cells containing piG_23 and piG_23b we were able to compare the toxic effect of mazF overexpression regarding length in individual cell lineages after 100 ng/mL DOX induction and without induction. We measured the cell size of about 25 to 40 cells using Fiji software (Figure 26). At time point 0, a similar cell length of about 2.2 µm was measured for all setups, except for non-induced piG_23b cells, which were slightly larger. This could, however, not be explained. After 3 hours, significantly smaller cells were observed for induced piG_23 samples with a size of 1.5 µm compared to non-induced cells (2.3 µm). There was no difference in cell length for cells containing piG_23b (induced and non-induced) at the same time point.
With this experiment, we demonstrated that by replacing the stronger RBS with the one of the riboswitch, induction with DOX leads to inhibited growth inhibition and cell death. Regarding the measurement of the cell length, the piG_23 samples that we used were not grown in M9 as the piG_23b samples, but in LB medium (previous experiment). Even if this means that the results are only limitedly comparable, we observed a difference in cell length for non-induced and induced piG_23 cells. We were now very interested to see if we would obtain the same toxic effect if we add the strong RBS to the riboswitch. Unfortunately this was out of the capabilities of our iGEM project.
Detection of Antitoxin Expression
Since we were able to finally clone the plasmid containing the antitoxin (piG_08), we wanted to determine antitoxin production under the constitutive AmpR promoter with western blotting.
The experiment was performed with a plasmid (piG_03). The piG_08 was transformed into E. coli MG1655. We then transferred the bacteria into flasks with 50 mL M9 medium and 34 mg/mL chloramphenicol. The cells grew at 37°C until an OD600 = 0.6 was reached. For Western Blot, 0.9 mL of culture were taken at each time point 0, 12 and 24 hours and centrifuged at 12000 rpm for 2 minutes.
Figure 28 shows the western blot results of the samples with cells containing piG_08 at time point 0, 12 and 24 hours after DOX induction. To detect mazE antitoxin on a western blot, we fused a his-tag at the N-term. MazE has a molecular weight of 9 kDa. By using a non-degrading buffer it is possible to determine whether the antitoxin is also present as a dimer (18 kDa) in the cell as it allows native protein folding to be preserved to some extent. Unfortunately, we observed no bands at 9 kDa in or 18 kDa for all three replicates (Figure 28).
One of the reasons that we can not see any antitoxin bands could be due to its fast degradation rate. MazE has a half life of around 30 min with an active degradation by the clpPA protease . Therefore, it is possible that the high degradation causes the overall concentration of mazE to be too low to be detected. Furthermore, since MazE is very small it can be difficult to detect it on SDS gels. We would have spent more time trying to optimize the expression and detection if we would have had more time. We therefore can not come to a conclusion whether the antitoxin was not produced at all or if it just got degraded before it could be seen on the gel.
-  Mager A, Safran T, Engelberg-Kulka H. Intracellular Localization of the Proteins Encoded by Some Type II Toxin-Antitoxin Systems in Escherichia coli. Davies JE, editor. mBio. 2021 Aug 31;12(4).
-  Espah Borujeni A, Mishler DM, Wang J, Huso W, Salis HM. Automated physics-based design of synthetic riboswitches from diverse RNA aptamers. Nucleic Acids Research. 2015 Nov 30;44(1):1–13.
-  Marianovsky I, Aizenman E, Engelberg-Kulka H, Glaser G. The Regulation of the Escherichia coli mazEF Promoter Involves an Unusual Alternating Palindrome. Journal of Biological Chemistry. 2001 Feb;276(8):5975–84.
-  Li GW, Burkhardt D, Gross C, Weissman Jonathan S. Quantifying Absolute Protein Synthesis Rates Reveals Principles Underlying Allocation of Cellular Resources. Cell. 2014 Apr;157(3):624–35.
-  Jimmy S, Saha CK, Kurata T, Stavropoulos C, Oliveira SRA, Koh A, et al. A widespread toxin−antitoxin system exploiting growth control via alarmone signaling. Proceedings of the National Academy of Sciences. 2020 Apr 28;117(19):10500–10.
-  Aizenman E, Engelberg-Kulka H, Glaser G. An Escherichia coli chromosomal “addiction module” regulated by guanosine [corrected] 3’,5’-bispyrophosphate: a model for programmed bacterial cell death.. Proceedings of the National Academy of Sciences. 1996 Jun 11;93(12):6059–63.