Overview
Through existing literature on combating invasive zebra and quagga mussels, we came across the bacterium P. protegens, which naturally produces an insecticidal toxin called FitD. FitD is encoded by the fitD gene and its transcription is activated by FitG. Both genes, fitD and fitG, are present in the genome of P. protegens. Scientific publications on a related Pseudomonas strain suggest that the FitD toxin is not only effective as an insecticide, but also toxic to Dreissena mussels (Daniel P.Molloy et al., 2013).
Our project consists in genetically modifying P. protegens to improve its production of the FitD toxin to kill the invasive quagga mussel effectively. When we genetically reprogrammed our bacteria, we applied engineering principles and underwent the following four stages: design, build, test, and learn.
1st cycle - Design
In order to increase our chances of success, we designed different strategies which independently aimed at overexpressing the FitD toxin in the bacterium P. protegens. We, indeed, thought of overexpressing the toxin by either (a) introducing simply an extra copy of its gene (fitD ) on a plasmid, or (b) introducing an extra copy of its activator (fitG ). Considering that expressing extra copies of these genes might be toxic or represent a significant metabolic burden on our cells, we designed and built vectors bearing both genes under the regulation of a constitutive or inducible promoter.
1st cycle – Build
To achieve the overexpression of the FitD toxin in P. protegens, we built four different plasmids containing either the fitD or fitG genes, which were amplified directly from the genome of the strain P. protegens CHA0 (see Experiments):
We registered the four plasmids with the numbers BBa_K4209011, BBa_K4209012, BBa_K4209013 and BBa_K4209014 in the registry for composite Parts.
We chose pSEVA plasmids as backbones for our cloning as they are designed to be compatible with Pseudomonas bacteria. Gibson assembly was used to clone either the fitD or fitG gene in our vector of choice (see Figure 1; see page Design , see page Experiments ). The inducible plasmids are based on the promoter lacIq-Ptrc, which IPTG can induce. The constitutive plasmids are based on the promoter EM7. All four constructs provide resistance to the antibiotic kanamycin.
The resulting plasmids were then introduced into competent E. coli DH5α cells by electroporation (see Experiments ) and cells were plated onto selective media.
1st cycle - Test
Our transformations resulted in colonies growing on our selective media. To check if we had obtained the expected plasmids, we performed colony PCR and visualized our PCR products by gel electrophoresis.
Results from our colony PCR showed DNA fragments of the expected sizes, indicating that we successfully cloned our desired plasmids (Figure 2).
Additionally, the PCR products were sent for sequencing to check for any potential mutations and the results obtained confirmed that our plasmid assemblies were successful (see Notebook).
Having built our plasmids of interest, we introduced them to P. protegens by electroporation. Unfortunately, our first trial to electroporate DNA into our cells was unsuccessful, as we could not obtain transformants on our selective media. (see Notebook)
1st cycle - Learn
We hypothetised that the failure of our attempt to electroporate DNA into P. protegens was due to low concentration of our plasmids. To circumvent this issue, we followed the protocol kindly provided by a laboratory technician from our department working with P. protegens CHA0. He recommended using 100 ng plasmid DNA for 50 uL of competent cells. We, therefore, measured our plasmid concentration with Nanodrop and obtained a concentration of 29.6 ng/uL. It allowed us to calculate a required adjusted volume of 3.4 uL to add to our cells for the transformation, but it also failed. We measured our DNA sample again with a Qubit assay for dsDNA to obtain a reliable concentration. The assay indicated an actual concentration of 11 ng/uL for our sample, which was significantly lower than our previous measurements. We then knew we needed more plasmid DNA for the transformation to be successful. (For more details see Notebook)
2nd cycle - Design and build
To obtain a higher concentration of plasmid DNA, we extracted plasmids from 9 ml of overnight liquid cultures instead of the previously 3 ml we had used. Using a speed-vac device, we also concentrated our sample to a final concentration of over 100 ng/ul by partially evaporating our sample under reduced pressure. (see Notebook)
Using the obtained concentrated DNA sample for transformation, we successfully transformed P. protegens and got transformant colonies on selective media . We learned that a very high concentration of plasmids was needed to transform successfully P. protegens. Having this information at hand, we performed all further transformations of these bacteria with highly concentrated plasmid preparations.
2nd cycle - Test
To check whether our plasmids were indeed transferred correctly into P. protegens, we performed colony PCRs and loaded the PCR products on a 1% gel. The PCR products were also sent for sequencing. We obtained amplicons of the expected size and sequencing results confirmed that the genes were correctly cloned into the plasmids (see Notebook).
At that point, we had successfully engineered all four different constructs of transformed bacteria and could proceed to lyse them and test their efficacy in killing mussels. Our subsequent experiments with mussels showed that our fitG construct kills significantly more mussels than the wild-type P. protegens CHA0 strain. Detailed information on bacteria lysis and mussel-testing can be found under Experiments and Measurement. The results of mussel-tests can be found under Proof of concept.
3rd cycle Design
We had successfully constructed 4 different strains whose toxicity we wished to test on mussels. In the next step, we needed to lyse the transformed bacteria for two reasons:
As we overexpress the protein FitD but not its transporters to the cellular membrane, some FitD is possibly not secreted and remains inside the cell, which may have less effect on the mussels (see Design). Lysing the bacteria would fix this problem because the cell content would be released in the solution.
Another reason for lysing the bacteria is that genetically modified organisms cannot be released into the environment. As we wanted to develop a product that is applicable in real-life, we wished to inactivate the bacteria after they produce the FitD toxin. The final product would contain dried lysed bacteria and FitD toxin and could be applied in mussel-infested pipes.
As we needed the FitD protein to stay stable, heat could not be considered as elimination method. Chemical treatments using alcohol, aldehydes, or chlorine cannot be considered because of their unknown effect on mussels and the environment. We could not use antibiotics either because of the antibiotic-resistance risks. After some research we found that sonication, a method based on applying ultrasonic frequencies to bacterial samples, would be efficient to lyse our bacteria without adversly affecting our protein (see Experiments).
3rd cycle Test and learn
We tested if the inactivation by sonication worked by serially diluting and plating our sonicated samples on selective media to count colony-forming units (CFUs). In the first sonication program, we used a pulse of 1s on/ 1s off during 30s. Unfortunately, there were still too many surviving bacteria. By increasing the sonication period up to 40s, we hoped to eliminate the bacteria effectively. We had to recognize that by doing so, the samples began to heat, and we ran the risk that the proteins got damaged.
4th cycle design
At that point, we needed to redesign our inactivation method and started looking for other techniques. We finally had the idea of using bead-beating: a mechanical way of killing bacteria by letting them collide with ceramic beads in a machine that stirs the sample.
4th cycle test
The plating of the bead-beater samples on restrictive media confirmed that this method was effective in lysing the bacteria: by microscopy we observed that 98% of our bacteria are lysed (see Experiments). Therefore, we used bead-beating for further experiments.
After the experiments with mussels, it is confirmed that one of our constructs, the fitG constitutive transformation, actually kills significantly more mussels than the wild-type strain of P. protegens CHA0 (see Proof of concept).
Overview
After extensive research on substances effective for combating mussels, we found that certain algae (e.g., eelgrass algae) naturally produce an anti-fouling substance called zosteric acid, which prevents the attachment of bivalves on surfaces. We, therefore, set out to modify bacteria to produce zosteric acid, a sulfate ester of the intermediate p-coumaric acid. (Jendersen et al.,2019)
When genetically modifying our bacteria, we followed engineering principles and underwent the following four stages: design, build, test, and learn.
1st cycle - Design
We based our design on the following pathway:
Production of zosteric acid can be achieved via a two-step enzymatic pathway. First, the tyrosine-ammonia lyase enzyme (TAL) forms p-coumaric acid by non-oxidative deamination of tyrosine. The sulfotransferase enzyme (PST) then converts p-coumaric acid to zosteric acid. We, therefore, wished to engineer the bacterium E. coli BL21 DE3 to express these two enzymes.
Since the production of zosteric acid requires much sulfate, we also aimed to modify our bacteria to increase uptake and activation of sulfate by overexpression of several enzymes, such as the Cys proteins. Ultimately, we designed our expression system as two plasmids: one catalyzing the transformation of tyrosine into zosteric acid (catalytic vector) and another plasmid for sulfate uptake (transport vector). The resulting plasmids were to be transformed into E. coli BL21 DE3, a strain with which zosteric acid has already been successfully produced (Jendersen et al.,2019).
1st cycle - Build
To achieve the expression of zosteric acid by E.coli BL21 DE3, we built four different plasmids. Two of them were catalytic plasmids, and two were transport plasmids.
To catalyze tyrosine into zosteric acid, two genes were required: tal–fjo and SULT1A1, encoding the TAL and PST enzymes, respectively. Since the gene topology may affect the production efficiency, we built two variant plasmids with different gene orders. The genes tal-fjo and SULT1A1 were synthesized and subsequently amplified by PCR (see experiments). We registered the tal-fjo biobrick under the number BBa_K4209000 in the registry for basic Parts.
For the transport plasmid, one gene was required for sulfate activation: the cysDNCQ gene that encodes the proteins CysDN, CysC and CysQ. An additional gene responsible for sulfate uptake was also required. For this, we designed two plasmids: one plasmid bearing the cysPUWA gene and another the cysP gene (corresponding to cysZ in the pathway) as the uptake gene. The cys genes were all amplified directly from the genome of Bacillus subtilis. (see Notebook). We registered the cysP biobrick under the number BBa_K4209004 in the registry for basic Parts.
By loading the PCR products on an agarose gel, we wanted to confirm whether the amplification of our genes was successful.
We obtained bands of expected sizes for all genes (for more details see Results).
For the assembly of these genes we needed backbones that were amplified by PCR from two different template plasmids, pCOLA and pET Duet (changed to pET17b later), chosen for their optimization for E. Coli and their IPTG inducible promoters and providing different antibiotic resistances. After amplification, the pET Duet backbone fragments needed to be purified by gel-purification and QIAGEN kit because of nonspecific primer binding. Using Gibson assembly technique, we finally cloned the amplified genes in the corresponding backbones.
1st cycle - Test
Subsequently, E. coli DH5α was transformed with our assembled plasmids using electroporation (see Experiments). Colony PCR and sequencing verification of obtained transformants confirmed the successful assembly of both transport plasmids: the genes cysDNCQ as well as cysP or cysPUWA were cloned in the plasmid without mutation (see Notebook). We registered both transport plasmids with the numbers BBa_K4209009 and BBa_K4209010 in the registry for composite Parts.
Unfortunately, no colony of the bacteria transformed with the catalytic plasmids grew on selective media.
1st cycle – Learn
After some electrophoresis tests, we found that the step of purification of the pETDuet backbone was responsible for the failure of the catalytic plasmids’ assembly: After gel-extraction and purification using QIAGEN kit, we performed electrophoresis which showed that after purification, no DNA was present in our sample. Despite multiple attempts, we were unable to perform the purification of the backbone fragments successfully. That is why we chose a new template plasmid for the backbone of the catalytic plasmids.
2nd cycle – Design
We changed the backbone for the pET17b plasmid which allowed specific primer binding for the amplification of the backbone. Therefore, subsequent gel-purification was not necessary anymore before assembling the plasmids. However, the pET17b template plasmid does not contain the domain for Lac-regulation of the catalytic genes, so we decided to obtain this domain from the template plasmid pAND-MCS, donated by Schaerli-lab, and clone it into the new backbone pET17b.
2nd cycle - Build
We performed PCR on the template pET17b to obtain the new backbone and on pAND-MCS to amplify the lacI and Ptac promoter region. Electrophoresis of the amplified fragments gave bands of the expected sizes:
With the new backbone pET17 and lac-regulation domain, we assembled the catalytic plasmids using Gibson-technique. Subsequently, we transformed E. coli DH5α with the different catalytic and transport plasmids. The transformed bacteria were incubated overnight on a selective media. The next day, we had colonies that grew on the plates (see Notebook).
2nd cycle – Test
After performing colony-PCR on our transformed bacteria, we ran an agarose gel to see whether we got bands of the expected size, which was the case. Sequencing results showed us that all genes were present in the plasmid (see Notebook). We registered both transport plasmids with the numbers BBa_K4209007 and BBa_K4209008 in the registry for composite Parts.
At this point, we successfully engineered our different transport and catalytic plasmids. In the next step, we had to incorporate the different combinations of plasmids into E. coli BL21 DE3. The results of these transformations can be found under Results.
References
- Daniel P. Molloy, Denise A. Mayer, Michael J. Gaylo, John T. Morse, Kathleen T. Presti a, Paul M. Sawyko, Alexander Y. Karatayev, Lyubov E. Burlakova, Franck Laruelle, Kimi C. Nishikawa, Barbara H. Griffin. (2013, January 5). Pseudomonas fluorescens strain CL145A – A biopesticide for the control of zebra and quagga mussels (Bivalvia: Dreissenidae). Journal of Invertebrate Pathology.
- Christian Bille Jendresen , Alex Toftgaard Nielsen. (6. September 2019). Production of zosteric acid and other sulfated phenolic biochemicals in microbial cell factories. Nat Commun 10.