Team:UZurich/Design
Design
The immune system of the plant kingdom contains a staggering number of receptors called pattern recognition receptors (PRR) which can detect an equally enormous number of epitopes which originate from a wide range of bacterial and fungal plant pathogens (see our PRR guide for more in-depth information). We were interested in a subsection of these receptors which are located at the plasma membrane and heterodimerize with the brassinosteroid insensitive 1-associated receptor kinase 1 (BAK1) upon exposure to a specific bacterial elicitor, which is usually a short peptide belonging to bacterial protein. This dimerization then leads to the interaction of the cytoplasmic kinase parts of the receptor and BAK1, initiating a signal cascade which then leads to the immune response, for example reactive oxygen species (ROS) production.
Our goal was to repurpose some of these receptors to be utilized as a biosensor which produces a visual output upon exposure to bacterial elicitors, which are peptide fragments of bacterial proteins. This could be used to approximate the total bacterial load of a water sample in a rapid and affordable manner which would deliver researchers a new toolkit for determining water quality.
Chassis
We chose Saccharomyces cervisiae (brewer’s yeast) and Chlamydomonas reinhardtii, a single cell green algae, as our chassis. We chose Saccharomyces cervisiae because it is one of the most used and well understood organisms in synthetic biology. Another appealing aspect of Saccharomyces cervisiae is it’s ability to be lyophilized (freeze-dried) and thus could be stored for a long time at room temperature, facilitating it’s usage as a biosensor.
But Saccharomyces cervisiae is a fungus and thus quite distantly related to the higher plants from which our PRRs originate, which might complicate the correct expression of said receptors. Thus we also wanted to try out something more closely related. Our choice fell on Chlamydomonas reinhardtii, which, as a green alga, belongs to the same kingdom as Arabidopsis thaliana and Solanum lycopersicum, Plantae, although it is of course a much simpler organism. Chlamydomonas reinhardtii is a popular organism for biosynthesis and has been used in previous iGEM projects, indicating its use as a potential chassis for our receptors.
Receptors
The first objective of our project was to determine which of the BAK1-dependent receptors we wanted to adapt into our system. To select our receptors two key criteria were used. First, they had to be well understood and characterized in plants. Second, they had to be able to recognize epitopes in bacteria that interest us. As PRRs are fundamentally plant immune system receptors they recognize not human but plant pathogens. However due to the extremely conserved nature of the epitopes they recognize, many of them can still detect a wide range of non plant pathogens. We determined the bacteria that interest us based on the WHO report on water safety [2]. Our sensor was to be able to recognize at least all harmful bacteria. To that end we performed many protein alignments in Jalview and compared them to the elicitor sequence of known PRRs [15, 16, 17].
Perfect conservation of the elf18 epitope (detected by EFR) across common waterborne bacteria, bacteria from WHO danger list in blue
Good conservation of fls15 (detected by FLS2) [15]
Moderate conservation of csp22 (detected by CORE) [16]
By this process we discovered our 3 receptors which, in theory, can recognize all bacteria on the WHO danger list. And many more, with EFR being able to recognize every known, tested, waterborne bacteria. This leads us to the very general sensitivity of our system. Additionally, this ensured we would not self recognize our own detector organisms or other nonbacterial microbe.
We tried to use the full-length receptors for our experiment, consisting of the epitope detecting extracellular ectodomain, the transmembrane domain which anchors the protein in the plasma membrane, and the cytoplasmic kinase domain. But the PCR amplification of said receptors of provided templates proved difficult and only worked for the BAK1 receptors. For the other receptors, we opted to have them synthesised by IDT. To save time and use less base pairs, we decided that we wouldn’t synthesise the kinase domains of the PRR, as they don’t seem to play a role in dimerization [1]. However, we left a small flexible part of the cytoplasmic part of the PRRs called the juxtamembrane domain as they might contribute to the stability of the receptor in the membrane, although this is just conjecture. All receptors from which we removed the kinase domain have an e at the beginning of their name, which stands for ectodomain (eBAK1 ,eEFR, eCORE, eFLS2).
When we ordered the receptors, we optimized the DNA sequence in such a way that they could be expressed in both Saccharomyces cervisiae and Chlamydomonas reinhardtii. These organisms have different codon usage biases. While Saccharomyces cervisiae is pretty tolerant in that aspect, Chlamydomonas reinhardtii is very picky and often has one prefered codon for each amino acid, while the other degenerate codons are very rarely found in the native genome. The presence of multiple rare codons in a row can slow down or even terminate the translation of our constructs, which is why we designed our constructs in such a way that there are the least amount of rare codons for both species. You can check out the different versions of each part on our Parts page.
EFR
BBa_K3610007
EFR (EF-Tu receptor) is the model example of a plant PRR as it is the most well-known and understood receptor of that family. It has been first discovered in Arabidopsis thaliana where it has been found to dimerize with BAK1 upon exposure to the elf18 peptide which is part of elongation factor thermal unstable (EF-Tu), an elongation factor found in almost all bacteria. It is also a very abundant protein, for example, it makes up up to 10 % of the total protein amount in Escherichia coli [3].
FLS2
BBa_K3610009
FLS2 recognises the flg22 and flg15 peptides which are part of bacterial flagellin, which, as the name implies, is an integral part of the bacterial flagellum which assists a lot of bacterial species in locomotion. There are 2 well characterised FLS2 variants in Arabidopsis thaliana and Solanum lycopersicum (tomato). We chose FLS2 from Solanum lycopersicum as this variant recognises the flg15 and flg22 peptides equally well, while the one from Arabidopsis thaliana has a much higher affinity for flg22 compared to flg15.
CORE
BBa_K3610010
The CORE receptor is a newer discovery in the family of plant PRR family compared to the two receptors above. It recognises the peptide csp22 peptide which is part of the cold shock protein (CSP) which regulates gene expression of bacteria as a reaction to cold temperatures. It is found in Solanum lycopersicum.
BAK1
BBa_K3610003
BRI1-associated receptor kinase (BAK1) is a receptor which was first discovered as part of brassinosteroid signaling in Arabidopsis thaliana, but has since been identified to play a central role in the sensing of bacterial elicitors. It acts as a binding partner for all the aforementioned receptors upon exposure to the specific elicitor which is necessary for the activation of the kinase situated at the cytoplasmic parts of the binding partners. Without BAK1, the EFR/FLS2/CORE could still bind the elicitor, but the kinase would remain inactive, inhibiting any immune reaction.
Protein structures of the ectodomains of the receptors we used, from left to right: EFR (Uniprot COLGT6), FLS2 (Uniprot Q9FL28), CORE (Uniprot Q9SD64), BAK1 (Uniprot Q94F62)
Signal peptides
To increase the likelihood of correct trafficking of our constructs to the cellular membrane, we replaced the native signal peptides with signal peptides specific to our chassis which are responsible for the excretion of proteins. As our constructs possess a transmembrane domain, they shouldn’t get excreted, but retained as a single-pass transmembrane domain. To test if replacing the signal peptide improves trafficking, we tested two different BAK1 constructs, one containing the native signal peptide, called BAK+ and one where we replaced the signal peptide, called BAK-.
Alpha factor
BBa_K2809002
We used the signal peptide and the pro-sequence of the mating factor alpha of yeast, which has been used to great success by other iGEM teams for cellular excretion. Both of these parts get cleaved off the constructs at different stages of cellular trafficking.
SP7
BBa_K3610012
For the constructs expressed in Chlamydomonas reinhardtii, we used the signal peptide SP7 which has been recently characterised in a paper by Molino et al. [4]
Reporters
HA and FLAG tag
We included common antibody-tags in our constructs, these being the HA tag, which is derived from the Human influenza surface glycoprotein Hemagglutinin, and the FLAG tag, an artificially designed tag. We attached the HA tag to the BAK1 receptor, while the FLAG tag was attached to the CORE, EFR and FLS2. By integrating these tags in our constructs, it is possible to detect interaction between BAK1 and CORE/EFR/FLS2 via co-immunoprecipitation (Co-IP).
YFP
Cellular trafficking is a complicated subject which involves multiple steps on a protein's journey from the ribosome to its target location, which in our case is the cell membrane, where it would be able to come into contact with the bacterial epitopes which we hope to detect. As the cellular trafficking varies from organism to organism, we have to make sure that our constructs arrive at the cellular membrane, as it would do us no good if the receptor got stuck in the endoplasmic reticulum ,a common issue with protein expression To that end, we attached a YFP protein to the C terminal end of our receptor parts. We can also use YFP as an antibody tag to measure the expression of Receptor-YFP fusions.
Split mCherry
We wanted to test if dimerization occurs between BAK1 and EFR/FLS2/CORE after exposure to the respective epitope by performing an Bimolecular fluorescence complementation (BiFC) assay using a split mCherry system.
BiFC works by taking a full length fluorescent protein which is then split into two parts. The two halves don’t exhibit any fluorescence on their own, but when brought into close proximity, they complement each other, reconstituting the full fluorescent protein which exhibits fluorescence. The main advantage of this system is that it doesn’t require any substrate to produce a visual output. But it also has a low signal to noise ratio, as structures in the cells can also be excited at the same wavelength as mCherry, producing what is called “autofluorescence”.
We attached the N-terminal half (called mCherry A) to the C-terminus of eBAK1 while we fused the C-terminal half (called mCherry B) to the ectodomain of EFR/FLS2/CORE. We used eBAK1 in the dimerisation experiments, as using the full lenght BAK1 would mean that the two parts of the BiFC system would be at different “heights”, which would make the interaction between these two parts impossible. We also included a linker between the juxtamembrane domain and the reporter parts consisting of the amino acid sequence (GGS)5. This was done to increase the freedom of movement for these parts to ensure that they could orient themselves in such a way that they could interact correctly.
Illustrations of part BBa_K3610040 (eEFR::mCherryB) and part BBa_K3610034 (eBAK1::mCherryA) as example for a set of constructs that we used to investigate dimerisation using the BiFC system
NanoBiT
NanoBiT® is a system developed by the Promega company in which they split their patented luciferase NanoLuc®. They split the luciferase into two pieces called Large BiT and Small BiT which, similarly to the BiFC system, complement each other when brought into proximity. The main advantage compared to the BiFC system is its much higher light intensity and better signal to noise ratio, which is due to the fact that it doesn’t require any excitation to emit light, but a substrate which in the case of NanoLuc® is Furimazine. This requirement for a substrate, which is quite expensive, is also the only drawback of this system which could limit its usage.
We used the parts already listed on the registry, but there are also other variants of this system which differ in their dissociation constant (Kd) as described by Dixon et al. [5]. If there is high background luminescence without the addition of any receptor elicitors, this means that the interaction of the NanoBiT® is driving the dimerization of our constructs, rather than our receptors. If this is the case, we could try out a variant in which the NanoBiT® parts interact less.
Illustrations of part BBa_K3610043 (eEFR::SmallBiT) and part BBa_K3610038 (eBAK1::LargeBiT) as example for a set of constructs that we used to investigate dimerisation using the NanoBiT® system
Vectors
Saccharomyces cerevisiae vectors
For expressing our constructs in yeast, we used a multi kingdom golden gate cloning platform developed by Chiasson et al. [6]. It allowed us to combine 6 elements into one expression cassette. To adapt our basic parts into this system, they need to be flanked by 2 pairs of IIS recognition sites which we integrated either via PCR with custom overhangs or during synthesis. The outer pair is a Esp3l cutting site, which after restriction digest by Esp3l leaves a 4 bp overhang that fits into an entry vector which was likewise cut with Esp3l (see figure A and B. Note that we used Esp3l and a vector containing Esp3l cutting sites instead of Bpil).
Entry cloning strategy. A: Overview of the primer components for amplifying a new basic part B: inserting the amplified part via digestion-ligation into the entry vector. From Chiasson et al. [6]
After we successfully inserted our receptor and detector parts into the entry vector, we are able to assemble them into one expression cassette (illustration below). The Primer overhangs contain a second recognition site for the BsaI enzyme. To ensure that each of the 6 parts gets assembled in the correct order, the overhang after BsaI restriction is different for each part, depending on the intended position of the part. The generalised order of our parts is:
| Position | A-B | B-C | C-D | D-E | E-F | F-G |
| Part | promoter | signal peptide | receptor | reporter | terminator | dummy |
The promoter, terminator and dummy parts were all part of the kit that was kindly provided by David Chiasson. The dummy parts are just short DNA fragments without any function, they just act as a placeholder if we don’t need any part at this position. This plasmid is then ready to be transformed into yeast cells.
Assembly of parts from level I entry vectors into a level II vector. adapted from Chiasson et al. [6]
The LII plasmids at our disposal differ in two important ways. The first difference was the way in which they replicated inside of the yeast cell. 2µ plasmids are present in high copy numbers inside a cell while centromeric plasmids behave like a chromosome and thus replicate in lower numbers. Integration vectors meanwhile integrate themselves into the yeast genome via homologous recombination, ensuring the establishment of stable transformed cell lines. We chose a centromeric plasmid, as 2µ plasmids could yield to high expression of our constructs which could prove toxic for the yeast cells. Integration vectors could be used once we get a working construct that we could distribute.
The other way in which the plasmids differ is in the selectable marker that they carry. Most of them carry an auxotrophic marker. This means that they give the yeast in which the plasmid is expressed the ability to to synthesise a nutrient which would otherwise need to be supplemented in the growth medium. Our strain for example lacks the ability to synthesise tryptophan, histidin, leucine and uracil. We chose the plasmid which allows our yeast to synthesise tryptophan after transformation. After transformation, we grew our yeast on agar plates containing so called synthetic defined (SD) medium which includes all previously named nutrients except tryptophan. Yeast which hasn’t been transformed wouldn’t be able to grow on these plates as they can’t synthesise the vital tryptophan, while transformed yeast would be able to grow.
But we don’t only want to express one construct in yeast, like in the case where we want to test the localisation of our constructs, but sometimes also two, for example when we want to test the dimerisation of BAK1 with EFR/FLS2/CORE. For that we need a second plasmid with another selectable marker. We chose one which enables yeast to express the protein kanamycin acyltransferase. This protein makes yeast resistant to the aminoglycoside G418 Geneticin. So if want to coexpress two receptors with split reporters, take BAK1::LargeBiT and EFR::SmallBiT for an example we would clone BAK1::LargeBiT into the plasmid with the tryptophan marker and EFR::SmallBiT into the plasmid containing the Geneticin marker. We would transform yeast with both of them and then plate them onto an agar plate containing SD -tryptophan, supplemented by Geneticin.
Chlamydomonas reinhardtii vectors
The vector that we use to express our constructs in Chlamydomonas reinhardtii, called pJP32, has been graciously provided by Joao Vitor Molino. Unlike the yeast plasmids, pJP32 doesn’t contain IIS restriction sites, but the classical restriction sites for XhoI and BamHI. This makes golden gate assembly of our basic parts into this plasmid impractical. Instead we pursued this strategy: We would clone our Chlamydomonas constructs first into the yeast plasmids via golden gate, and then PCR amplify the parts that we want to be translated in Chlamydomonas reinhardtii. The signal peptide SP7 is already included in the plasmid, so we don’t need to include it during assembly of our constructs. To get the fragments into pJP32, we used a technique called In-Fusion® cloning developed by Takara bio. This technique relies on the existence of 5’ and 3’ ends the size of 15 bp in the Fragment which are homologous to the sticky ends of the plasmid after restriction by XhoI and BamHI. The exonuclease in the In-Fusion reaction mixture chews back the 3’->5’ ends of the vector and the fragment. This creates overhangs which allow the fragment and the vector to anneal. A DNA-polymerase then fills the remaining gap and a DNA ligase ligates the nicks.
The plasmid is quite unique in the way it expresses its selection marker, the ble gene which confers resistance to the antibiotic zeocin. Instead of being regulated by a separate promotor, it gets expressed in the same ORF as our target cloned receptor constructs. To ensure that this doesn’t result in a fusion protein, F2A, a self-cleaving peptide, has been placed between these two. During translation, the specific amino acid sequence of F2A causes the peptide bond to break at a specific site, leading to the separation of our construct and the zeocin resistance protein.
But what did we plan to do when we wanted to express two constructs at the same time, like with eBAK::mchA and eEFR::mchB? We only had one chlamydomonas plasmid with one selectable marker, so we couldn’t express the constructs on two different plasmids. To solve this problem, we cloned both constructs in the same ORF, but used a special version of eBAK, which has been codon optimized for chlamydomonas. It contains the F2A peptide on the N-terminus, followed by the signal peptide SP7. When the constructs get translated, F2A causes them to separate and as both constructs contain the SP7 peptide, they should both get trafficked to the cell wall.
eEFR::mCherryB and eBAK::mCherryA in the pJP32 as an example of combining two constructs in one vector. Note the F2A peptide separating the two constructs. Plasmid illustration was made by SnapGene.
Experimental Design
Cloning
As previously described, for our cloning strategy Golden Gate Cloning [7] was the obvious choice.
We inserted the single fragments into our LI plasmids with a gentamicin 3’-acetyltransferase gene, a gene commonly used for selective survival, and amplified them in E. coli (DH10α) [8,9]. The bacterial cells were then spread on a plate containing gentamicin to select for cells that had taken up the plasmid.
In the next step we assembled the six fragments in a plasmid with a spectinomycin acetyltransferase and amplified the plasmids again in the same E. coli strain [9].
For expression in C. reinhardtii, the initial workflow was identical to expression in S. cerevisiae. The fragments were cloned into LII vectors in the same manner and then inserted into our chlamydomonas vector.
Whenever fragments had been cloned into a plasmid, the success of the reaction was confirmed by sequencing the plasmids to ensure the right fragments were inserted into the vectors and no mutations changing the protein have occurred.
After having prepared all the constructs on the correct plasmids, our chassis were transfected with the plasmids. As already established, when coexpressing constructs in S. cerevisiae, two different vectors were used.
The transformation of the yeast cells with two plasmids can be done simultaneously, in which case the organisms needed to be plated directly on the plate with both selection markers, or by sequential transfection. In our case, we tried to transform the yeast with both plasmids at once, but also set up a transformation with only one plasmid, as double-transformation has a higher propensity to failure.
For our second chassis C. reinhardtii, multiple constructs were inserted into one plasmid, which relieved us from dealing with multiple transformation steps.
Quantification of fluorescence levels
Since S. cerevisiae is an organism that has the possibility to autofluoresce, observing fluorescence with a confocal microscope is not enough to confirm expression. Fluorescence must additionally be detected and quantified with a fluorometric plate reader to compare the untreated yeast cells with the ones that took up the plasmids. During this experiment, the fluorescence intensities and the optical density (OD600) was measured of each sample. Additionally, a blank measurement was performed as a zero reference. The results were standardized for OD600, a step which was necessary as samples with higher OD values will naturally give higher measurement values when examined with the fluorescent plate reader.
In addition to assessing fluorescence intensity with the fluorometric plate reader, we were able to perform a flow cytometry experiment, which allowed us to measure fluorescence intensity with a different method. It was important to us to quantify the fluorescence with multiple different approaches as this increases the information coverage and allows us to draw a more complete picture in characterizing our parts. With flow cytometry, we measured the fluorescence intensities (488/530 FITC channel) of different biological replicates. Subsequently, the samples were pooled together and examined with the same method as well.
Avoiding Autofluorescence
Multiple components in a S. cerevisiae cell have the ability to autofluoresce, like tryptophan, pyridoxine and riboflavin. These side effects increase the unreliability of fluorescent microscopy results. Therefore, we aimed at minimizing these side effects by analyzing the cells in a synthetic defined medium instead of yeast extract peptone dextrose (YPD). The dramatic effect can be seen on the right: The two tubes in the upper left corner contain only medium without S. cerevisiae. The left-most tube contains pure YPD medium, which has extremely strong autofluoresence, while the synthetic YNB medium to the right of it shows almost no autofluorescence. We had many problem while conducting our experiments until we discovered this effect.
Confocal microscopy
In order to visualize expression and localization at the cell membrane of the full length receptors and the receptor ectodomains they were fused to a yellow-fluorescent protein with an excitation wavelength of 514 nm and an emission wavelength of 527 nm.
The cell membrane was stained with fm4-64, which fluoresces strongly after binding to the cell membrane (λEM = 640nm). The binding of the dye is happening rapidly and it is also reversible. If the time spent between staining and imaging is too long, then the dye will be taken up by the organism and stored in the vacuole [11].
Imaging with a confocal microscope for YFP and the fm4-64 stain shows the spatial overlap of the red fluorescence of the stain and the yellow fluorescence of the protein fused to the receptors.
Test for dimerization
The next crucial step was to see whether the receptors actually dimerize in the presence of the bacterial epitope. Different approaches to investigate this question were evaluated and finally, we chose to conduct experiments to measure the luminescent output of the split-reporter proteins and to additionally test for protein-protein interaction with co-immunoprecipitation.
Luminescence assay
The yeast cells, which were previously co-transformed with EFR and the coreceptor BAK1, were diluted to adjust the samples to specific optical densities, the same principle was applied to untransformed cells. Such a step was necessary to prevent higher luminescence intensity in one sample occurring merely through a higher cell density.
There are a few factors that need to be taken into consideration when doing dimerization assays. Since the interaction of the plant PRRs is ligand-dependent, dimerization should only be initiated when the bacterial MAMPs are present. This means testing for dimerization needs to be done with the bacterial peptides to which the receptors are specific and a control without the peptide. It must further be clarified whether the epitope does not elicit a response in cells which have not been transformed with the plasmids.
As mentioned previously, autofluorescence levels of yeast cells are high in YPD medium, which is why we switched to a synthetic defined medium for dilutions in this assay as well.
For the actual assay, 50μl of Nano-Glo solution (50:1 buffer to substrate) was added to a 50μl sample. The signal is then detected and quantified with a luminometer.
Co-Immunoprecipitation
Immunoprecipitation (IP) is a popular method to detect and also purify proteins from a sample using antibodies (AB) and an antibody-binding protein which is immobilized to a beaded support. The AB binds to a protein of interest and is precipitated by the antibody-binding protein. Proteins that are not captured with the beads can be washed away, after which the antigen can get eluted from the capturing-complex and analyzed in a Western Blot [14].
In Co-Immunoprecipitation, there is very little about the procedure that needs to be changed. After washing away the proteins which are not captured by the beads, one can choose to focus on the examination of the primary AB-target, which is the antigen. If a protein binds to another structure, the whole complex is precipitated by the antibody-binding protein, which leaves us with a sample containing both, the primary protein and its interaction partner, so one may choose to examine the secondary target [8].
Co-Immunoprecipitation has been widely used to identify protein-protein interaction. Since our system relies on protein-protein interaction as well, we decided to examine the epitope-induced dimerization with co-immunoprecipitation in addition to just measuring fluorescence or luminescence intensity levels.
All our constructs with the receptors fused to mCherry or NanoLuc contained either the FLAG-tag or a hemagglutinin(HA)-tag. The target-receptor (e.g. EFR) was fused to the SmallBit part of the NanoLuc protein and, therefore, contained the FLAG-tag in its sequence, while the coreceptor BAK1 was fused to LargeBit with a HA-tag. Should the receptors dimerize upon addition of the respective bacterial elicitor, immunoprecipitation for the FLAG-tag should also capture the BAK1 protein as it will attach itself to the AB-bead via the target-receptor.
After washing of the sample and elution of the proteins from the beads, the target receptor construct and the BAK1 construct can be identified in a Western Blot by using AB-staining of FLAG and HA.
Split NanoLuc Part Characterization & Improvement
In order to test the S. cerevisiae codon-optimized NanoBit system, the NanoLuc parts were fused to the FK506-binding protein (FKBP) and the FKBP–rapamycin binding (FRB) domain. These two proteins dimerize upon addition of rapamycin, resulting in the reconstitution of the split NanoLuc. For this assembly, Golden Gate Cloning was our cloning strategy of choice and the cloning workflow followed the same principle as for the construction of our receptor constructs.
However, we were faced with an unexpected predicament. The actual 15 amino acid long linker connecting the SmallBit and the rapamycin ligand turned out to be erroneous, meaning it was 5 amino acids shorter than designed, making it harder for the two parts to dimerize.
After some consideration, we decided to express the original SmallBit sequence in the yeast cells together with the codon optimized LargeBit, as SmallBit is a very short fragment which naturally does not contain any codons with an occurrence rate of only 10% or less in S. cerevisiae.
After successful transfection of our cells with either the constructs containing the original NanoBit sequences or with the constructs containing the codon optimized version of the LargeBit, a dimerization assay was performed in the same manner as previously described, with the only difference being the addition of rapamycin to induce dimerization of the NanoBits driven by the rapamycin-binding ligands.
[1] Albert M, Jehle AK, Mueller K, Eisele C, Lipschis M, Felix G. Arabidopsis thaliana pattern recognition receptors for bacterial elongation factor Tu and flagellin can be combined to form functional chimeric receptors. J Biol Chem. 2010;285:19035–19042.
[2] WHO: Guidelines for drinking-water quality, 4th edition, incorporating the 1st addendum, 2017, https://www.who.int/publications/i/item/9789241549950
[3] Weijland, A., Harmark, K., Cool, R.H., Anborgh, P.H. and Parmeggiani, A., 1992. Elongation factor Tu: a molecular switch in protein biosynthesis. Molecular microbiology, 6(6), pp.683-688.
[4] Molino, J.V.D., de Carvalho, J.C.M. and Mayfield, S.P., 2018. Comparison of secretory signal peptides for heterologous protein expression in microalgae: Expanding the secretion portfolio for Chlamydomonas reinhardtii. PloS one, 13(2), p.e0192433.]
[5] Dixon, A.S., Schwinn, M.K., Hall, M.P., Zimmerman, K., Otto, P., Lubben, T.H., Butler, B.L., Binkowski, B.F., Machleidt, T., Kirkland, T.A. and Wood, M.G., 2016. NanoLuc complementation reporter optimized for accurate measurement of protein interactions in cells. ACS chemical biology, 11(2), pp.400-408
[6] Chiasson, D., Giménez-Oya, V., Bircheneder, M., Bachmaier, S., Studtrucker, T., Ryan, J., Sollweck, K., Leonhardt, H., Boshart, M., Dietrich, P. and Parniske, M., 2019. A unified multi-kingdom Golden Gate cloning platform. Scientific reports, 9(1), pp.1-12.
[7]Engler, C., & Marillonnet, S. (2014). Golden Gate cloning. Methods in molecular biology (Clifton, N.J.), 1116, 119–131. https://doi.org/10.1007/978-1-62703-764-8_9
[8]Schwocho LR, Schaffner CP, Miller GH, Hare RS, Shaw KJ. Cloning and characterization of a 3-N-aminoglycoside acetyltransferase gene, aac(3)-Ib, from Pseudomonas aeruginosa. Antimicrob Agents Chemother. 1995;39(8):1790-1796. doi:10.1128/aac.39.8.1790
[9]Vardanyan, R. S., & Hruby, V. J. (2006). 32 - Antibiotics. In R. S. Vardanyan & V. J. Hruby (Eds.), Synthesis of Essential Drugs (pp. 425–498). Elsevier. https://doi.org/10.1016/B978-044452166-8/50032-7
[10]Park, J., Throop, A. L., & LaBaer, J. (2015). Site-specific recombinational cloning using gateway and in-fusion cloning schemes. Current protocols in molecular biology, 110, 3.20.1–3.20.23. https://doi.org/10.1002/0471142727.mb0320s110
[11]Rigal, A., Doyle, S. M., & Robert, S. (2015). Live cell imaging of FM4-64, a tool for tracing the endocytic pathways in Arabidopsis root cells. Methods in molecular biology (Clifton, N.J.), 1242, 93–103. https://doi.org/10.1007/978-1-4939-1902-4_9
[12]Mahmood, T., & Yang, P. C. (2012). Western blot: technique, theory, and trouble shooting. North American journal of medical sciences, 4(9), 429–434. https://doi.org/10.4103/1947-2714.100998
[13]Garrett, W. S., & Mellman, I. (2001). CHAPTER 16 - Studies of endocytosis. In M. T. Lotze & A. W. Thomson (Eds.), Dendritic Cells (Second Edition) (213-cp1). Academic Press. https://doi.org/10.1016/B978-012455851-9/50055-9
[14]Phizicky, E. M., & Fields, S. (1995). Protein-protein interactions: methods for detection and analysis. Microbiological reviews, 59(1), 94–123.
[15]Meindl, Thomas & Boller, Thomas & Felix, Georg. (2000). The Bacterial Elicitor Flagellin Activates Its Receptor in Tomato Cells According to the Address-Message Concept. The Plant cell. 12. 1783-94. 10.2307/3871189.
[16]Wei, Y., Caceres-Moreno, C., Jimenez-Gongora, T., Wang, K., Sang, Y., Lozano-Duran, R., & Macho, A. P. (2018). The Ralstonia solanacearum csp22 peptide, but not flagellin-derived peptides, is perceived by plants from the Solanaceae family. Plant biotechnology journal, 16(7), 1349–1362. https://doi.org/10.1111/pbi.12874
[17]Gernot Kunze, Cyril Zipfel, Silke Robatzek, Karsten Niehaus, Thomas Boller, Georg Felix, The Plant Cell Dec 2004, 16 (12) 3496-3507; DOI: 10.1105/tpc.104.026765