Every project starts with an idea. To bring the synthetic biology idea to life, a systematic approach is required. The process an engineer goes through in the development of a product can be divided into iterative stages. Each iteration will improve the system, based on knowledge gained from previous iterations. The four stages are Design, Build, Test, and Learn.
The Design stage is the first step in the engineering cycle. Firstly, the desired function should be defined. Secondly, a conceptual draft should be sketched, so that it can be questioned, tested, and simulated using various computational tools. After finding a successful design that, on paper, achieves the desired function, it is time to move on to the build stage.
The Build stage is where the designed construct is transformed into an organism of choice. The methods of assembly used in the build stage should consider the design specifics. This stage also encompasses the experimental methods in the first set of cycles discussed.
In the Test stage, the function of the modified organism, or the assembled construct inside the chassis, is evaluated.
Lastly in the Learn stage, the results from the test stage are used to improve the design. The initial design is rarely sufficient, so new ideas for the design are generated - hence the cyclic nature of the process. After adequate iterations, a useful design should succeed.
Our aim in the laboratory is to successfully degrade PFAS in water samples, and then later be able to utilize our solution to clean drinking and wastewater at large scale. During our project, we went through the engineering cycle multiple times. Below you can see our work optimizing our dehalogenases (DeHa) to better degrade PFAS, using PFOA as a model molecule. You can also read about our work making E. coli Rosetta more fluoride tolerant.
The specifics of how our final designed construct works and how it is assembled are described in greater detail on the Parts Registry page.
We designed a sequence where the dehalogenases (DeHa) could be expressed in a system, where higher protein efficiency results in higher antibiotic resistance. Kanamycin was the antibiotic we chose as a selection marker. See figure 2 for the sequence.
The initial construct was designed with a site occupied by an RFP (red fluorescent protein) gene. This gene would allow for differentiation between colonies that have been transformed and those that have not. This RFP site was a placeholder for the DeHa gene. The DeHa gene would be assembled by Golden Gate Cloning using PaqCI restriction enzyme, removing the RFP gene.
The construct contains a riboswitch associated with inducible kanamycin resistance. The riboswitch becomes active when bound by fluoride.
The assembly was simulated in Benchling and found to be successful.
We tried ordering our insertion piece in one part, but it was not able to be synthesized. Therefore, from now on, we ordered the insertion piece in two parts: the insertion sequence and the riboswitch part. This improved the quality and success of synthesis. The insertion piece is then made up of the two pieces, and then combined with Gibson assembly when inserted into the plasmid.
Using Gibson assembly, the insertion sequence was cloned into the pET-51b(+) expression vector and transformed into NEB competent E. coli. The DeHa gene was inserted using Golden Gate Cloning. Later, the plasmid was purified and transformed into our ALE1 Rosetta strain from our fluoride tolerance experiments.
To confirm the successful integration of our DeHa genes into BBa_K4868001, our insertion piece, we ran gel electrophoresis.
To test whether the riboswitch worked, we investigated the ability of the ALE1 strain to grow on plates with kanamycin, sodium fluoride (NaF), and Isopropyl β- d-1-thiogalactopyranoside (IPTG).
The expected results would be growth only when NaF was present, as this would result in expression of kanamycin resistance. It was found that there indeed was growth, but we also found growth with no NaF present, which was not expected.
We verified the correct assembly of BBa_K4868001 and the DeHas since the size of the bands matched the size of BBa_K4868001, as seen on figure 3a, and the DeHa genes, as seen on figure 3b.
We learned that it was difficult to differentiate some of the dehalogenases from RFP on the gel. This is because the RFP genes had a length of 675 bp. The genes encoding the DeHa enzymes range from 672 bp to 900 bp and therefore the bands on the gel from the DeHas and the RFP would be too close together to distinguish from each other.
We found that the bacteria were resistant to kanamycin when exposed to NaF, even though there was no riboswitch activation. This indicated a leaky riboswitch.
We also learned that using kanamycin resistance as a means of selection was suboptimal. It would be hard to determine whenever the bacteria were resistant against Kanamycin because of the riboswitch activation, or because the riboswitch was leaky. This would make it difficult to select the most effective enzyme since it was not possible to determine what caused the resistance.
After the previous iteration, multiple changes had to occur. Firstly, the selection marker associated with the riboswitch had to be changed from kanamycin resistance. Secondly, the riboswitch had to be reinvestigated to ensure proper activation.
For the selection marker, we decided to opt for super-folder green fluorescent protein (sfGFP), as this would allow for quantitative selection of the colonies.
We reevaluated the genetic elements surrounding the riboswitch, as they may be essential to the function. We found that in the paper “Point-of-Use Detection of Environmental Fluoride via a Cell-Free Riboswitch-Based Biosensor” by Thavarajah et al., they described the riboswitch1. Here they used a specific constitutive promoter, the Anderson Promoter (BBa_J23119_Spe1). We replaced the T7 promoter with an Anderson promoter to ensure the riboswitch would only be active when induced by fluoride.
Our insertion piece was further modified. From the Thavarajah et al. paper we also inserted the T1 terminator. We called this insertion piece BBa_K4868001. See the modified insertion piece in figure 4.
We cloned the new solution BBa_K4868001, into the pET-51b(+) expression vector using Gibson assembly. The assembled plasmids were then transformed into NEB competent cells. Following this procedure, the plasmids were purified and introduced into the ALE1 Rosetta strain from our fluoride tolerant experiments and the original E. coli Rosetta strain.
For the first test of the riboswitch, fluorescence microscopy was performed. We did this to test the fluorescence response of BBa_K4868001 to different concentrations of NaF.
The test was performed where the original Rosetta, ALE1 strains, and NEB competent stable cells were incubated overnight. The following day the bacterial cultures were grown in tubes with Ampicillin (100 mM), IPTG (0.4 mM), and varying amounts of NaF. The cultures were then fixated and examined under a fluorescence microscope to investigate which concentrations of NaF resulted in the highest sfGFP detection. The results can be seen in figure 5.
In addition to fluorescence microscopy, the samples were added to two 96-well plates with Ampicillin and varying amounts of NaF: one plate for OD600 measurements and one for fluorescence measurement. The samples in the 96-well plates were measured in a plate reader, as seen on figure 6, and Fluorescence-activated Cell Sorting (FACS) machine seen on figure 7. The first results obtained showed a fluorescence optimum at 100 mM NaF.
Further experiments were then conducted to narrow down the exact amount of NaF needed to be added for optimal effect. The added NaF to cultures were of the concentrations: 75 mM, 85 mM, 95 mM, 105 mM, 115 mM, and 125 mM.
The riboswitch was also tested using FACS and a microscope. At the FACS there was sometimes detected 2 population, 1 population showing fluorescence and 1 population without fluorescence. This indicates that the plasmid might be lost in some of the cells, which might be due to unstable ampicillin that loses efficiency after too many hours in liquid at 27-37°C. Upon investigating the samples under the fluorescence microscope, it was confirmed that some of the cells had lost their pladsmid, since they showed no fluorescence.
As seen in figure 5 the ALE1 strain did not fluoresce noticeably, no matter how high a concentration of fluoride it was exposed to. This was unfortunate since this indicated that the riboswitch wasn’t working. One explanation for the fluoride tolerance could be that the bacteria had improved their fluoride ion excretion. The fluoride ions could thereby not activate the riboswitch, which would result in the cells not fluorescing. We were excited to see that this was not the case for the original Rosetta strain and the NEB competent cells. This verified that the fluoride ions usually enter the cell, thereby activating the riboswitch.
FACS and plate reader results from figure 6 and 7 show that some of the DeHas first release fluorescence detectable at a NaF concentration higher than 50 mM. This indicates that the riboswitch has a low sensitivity, which is a problem since we in the lab can only dissolve PFOA at a 1.2 mM.
To accommodate the fluoride-toxicity challenge from the beginning of our project, we designed an adaptive laboratory evolution (ALE) experiment to confer a higher fluoride tolerance to our original E. coli Rosetta strain. We planned on inserting our pET-51b(+) expression vector with one of the DeHa genes, into this ALE strain.
For this purpose, we decided to use increasing concentrations of fluoride (using dissolved sodium fluoride (NaF)) and UV light to increase the fluoride tolerance of the bacteria. We decided that a 96-well plate would be suitable for this experiment because it allowed us to easily make replicates and dilution series of different NaF concentrations and the different lengths of UV exposure. We used a plate reader to measure the OD600 of the 96-well plates to provide insights into the bacterial growth efficacy of the different samples.
The original E. coil strain contained a plasmid with chloramphenicol resistance. We did not include chloramphenicol in the media to reduce growth inhibiting factors in the ALE experiment.
Because there is no DNA construct designed to transform into the chassis, this experiment did not have a build phase.
Each day, a new 96-well plate with varying NaF concentrations and bacteria selected from the most promising well of the 96-well plate experiment conducted the previous day. was incubated and OD600 was measured. The bacteria from the most promising well were then selected and used for further evolution.
The testing stage was repeated every day for a month.
To conclude the series of experiments, we tested whether there was a difference in fluoride tolerance between the original E. coli Rosetta strain and the Rosetta strain developed using ALE (from now on referred to as ALE1) in liquid media. In figures 8 and 9 the results from the comparison of original Rosetta and ALE1 can be seen.
The Rosetta strain had chloramphenicol resistance encoded on a plasmid at the beginning of our experiments. We learned that after a month of not having been exposed to chloramphenicol, the bacteria had lost the plasmid. Therefore, they were not resistant to chloramphenicol anymore. This was a problem because we wanted to select our bacteria, making sure they were not contaminated. This problem was fixed by purifying the plasmid from the original Rosetta strain and then reintroducing it into the ALE1 strain.
From the experiments performed in liquid media, a difference between the two strains is observed. The ALE1 strain survives in concentrations >225 mM NaF while the original Rosetta does not. In contrast to the liquid media, no difference was seen between the two strains when grown on agar plates containing NaF.
We decided to reevaluate our experimental design due to the complications from the previous iteration.
For this version of the ALE experiment, we exchanged the 96-well plate for LB agar plates. This would, however, limit the number of different concentrations we could test every day. Furthermore, we were limited by the level of precision in our laboratory scale since we had to measure NaF on a daily basis. The NaF concentration needed the next day was based on the results of the previous day. We hoped that this approach would yield more consistent results.
We started this experiment while still trying to make ALE 1 work. Therefore, we decided it would be best to start over with the original E. coli Rosetta strain, rather than trying to evolve ALE1 further due to the loss of chloramphenicol resistance.
Because there is no DNA construct designed to transform into the chassis, this experiment did not have a build phase.
The ability of the bacteria to grow on different concentrations of NaF was assessed by looking at the plates after incubating. The plate with the highest concentration of NaF that contained bacterial growth would be chosen and the bacteria would be radiated with UV light, resuspended, and plated on a new NaF containing agar plate.
Agar plates were made with chloramphenicol and different concentrations of NaF. The first time this was done, bacteria from an E. coli Rosetta overnight culture were used, but as the experiment went on, bacteria from the plate with the highest concentration of NaF that contained bacterial growth were used. When the plates were dry, they were exposed to UV light, sealed with parafilm, and left to incubate overnight.
After incubation, the plates were assessed and the plate containing the most colonies would be used for the next plates.
When the development reached a plateau, it was determined in an experiment whether the resulting strain from this experiment (which will be referred to as ALE2) was more fluoride tolerant than the original Rosetta strain and the ALE1 strain.
When examining the plates, it appeared as if the evolved ALE2 strain had improved fluoride tolerance. Comparison between the ALE2 strain to both the ALE1 strain and the Rosetta strain showed, that the ALE2 strain was more tolerant to fluoride than both the ALE1 strain and the original Rosetta strain which can be seen in figure 10. Here ALE2 grew on higher concentrations of NaF than ALE1 and the original Rosetta strain.
We used ALE1 and ALE2 for multiple of our other experiments but found that the fluoride tolerance for ALE1 and ALE2 was probably due to the bacteria pumping the fluoride out and therefore making our riboswitch inactive.
If you are interested in the Whole Genome Sequencing results for the ALE1 and ALE2 strains, please contact: sofielmoebes@gmail.com