Introduction

The goal of our iGEM project at the Rijksuniversiteit Groningen, is to genetically engineer nanobodies against avian influenza that are secreted into lung mucus by native chicken bacteria. To enable this we need to be able to express the nanobodies, secrete them into chicken lungs, and ensure the biosafety by containing its spread during the administration of the therapeutic. The following sections describe engineering details to achieve these goals.

Organism Selection

To limit microbiome disruption, we require the species of bacteria that expresses and secretes nanobodies to be native to the lungs of chicken. Some other key considerations during selection is its prevalence, survivability, frequency of escape, and potential risks to the lungs of chicken. An added advantage would be if the chosen organism has previously been genetically engineered. A study by Ngunjiri et al. [1], reported the lung microbial species composition of chicken from various age-groups. They did this by extracting the DNA of microorganisms sampled from the trachea, Ileum, nasal and cecum tissues of the lung. This DNA was then sequenced using primers for highly-variable regions of the conserved 16S rRNA, allowing for the identification of present species. Relative abundances of the species could then be determined, resulting in figure 1 below.

Figure 1. Dynamics of respiratory microbiota across age groups.

This data (figure 1) showed that while chicken lung microbiomes are highly dynamic, there are some genuses of microorganisms that are quite ubiquitous. For example, various lactobacillus species seed all four tissues of the lungs of younger chicken, and remain there quite long as well. In the study this was analyzed in more depth with the same method, resulting in figure 2 below.

Figure 2. Ecology of Lactobacillus in URT and LT sites. Bar charts show changes in relative abundance of genus-level bacterial taxa (vertical axis) with chicken layer age (horizontal axis). 01W, 03W etc. indicate ages of birds in weeks. Nasal, nasal cavity.

As can be seen, various species such as L. crispatus, L. gasseri, L. reuteri, L. salivarius, and L. vaginalis are found long-term in all 4 analyzed regions of the lungs, and are therefore good candidates for nanobody expression.

Beyond the ability to colonize the lungs, the ease of modification (including precedence, so whether there are existing protocols), effectiveness of (nanobody) protein expression, and its effects on chicken health are also important for our purposes. Further literary research showed that L. reuteri specifically was an ideal candidate, as it already fulfills a beneficial role in lung microbiome health through its production of reuterin, a antimicrobial molecule which protects lung mucus from foreign bacteria. Moreover, the organism has been previously used to express and secrete nanobodies already, like in a study by Krüger et al. [2], where nanobodies against Streptococcus mutans were expressed and secreted using L. reuteri for the purpose of in-situ passive immunization against the pathogen . This was in essence the same principle as we intended to do with avian influenza, hence we adopted some of the methods used in these studies to allow for streamlined modification of L. reuteri.

Cloning method: Golden Gate Assembly

The construction of assemblies described below was performed through Golden Gate Assembly. To use this method on a region of DNA a set of primers designed containing 3’ annealing regions complementary to the beginning and end of the target region (as with any PCR), but also contain a 5’ tail region containing a 3’ pointing type IIS restriction site. When these primers are used, the tails create 5’ and 3’ inward-pointing restriction sites which, when cut with the restriction enzyme, cleave off the restriction recognition sites. When an insert region and a vector are PCRed using this method and combined in a one-pot mixture with the restriction enzyme and ligase, the two fragments can be assembled in a scarless manner. In a PCR machine the restriction enzyme will first cut off the restriction recognition sites of both fragments, producing sticky-ends compatible with one-another as well as with the small cut off recognition site. The temperature is changed allowing for ligase to function and ligate the insert and vector fragment, or the fragments with the cut-off recognition sites. If the latter occurs it will be possible for the restriction enzyme to function again in the next round, though if the former occurs the restriction recognition site is no longer present and the final construct will have been assembled. Through many rounds this will result in the specific assembly of the added fragments in one pot.

This pot can then be immediately transformed and plated on selection plates. This process is illustrated below (figure 3).

Figure 3. Specific assembly of the added fragments in one pot.

Within this project, this method has been adapted to assemble all constructs, using the type IIS restriction enzyme BsaI, capable of cutting as follows: GGTCTC(1/5), thus generating a 4 bp overhang. Though a number of Golden Gate components were synthesized de novo, through the IDT DNA synthesis service [4], a number were PCRed using primers designed to insert BsaI restriction recognition sequences pointing inwards at the ends of the DNA fragments. The overhangs formed by these restriction enzymes vary per component, but have been standardized to allow for modularity amongst many of the constructs, as will be described in more detail in later sections.

Nanobody expression and secretion

In a stepwise fashion, first a fitting organism must be selected, followed by an investigation of how this organism could secrete small proteins like nanobodies. Then this protein secretion mechanism must be adapted to the nanobodies in (preferably) one plasmid, which can then be transformed into the selected organism.

In practice, this would result in the general assembly of: Vector, promoter, nanobody with fused tags, terminator. Though the organization of this depends heavily on precedence, allowing for more effective comparative analyses.

Once the construct is built an investigation into the expression and secretion efficiency of the nanobody must then follow, as well as the elucidation of the effectiveness of nanobody binding to the desired target. Both are also to be compared to an easily analyzed control nanobody. With the data acquired from the respective experiments it will be possible to determine the efficacy and feasibility of the nanobody-based therapeutic probiotic strain we hope to create.

Design considerations

Vector selection

Once a fitting host organism was selected it was a matter of finding past research groups which had worked with L. reuteri to express and secrete nanobodies. This was found in a study by Lizier et al. [5], where a comparative analysis between different expression vectors was performed on various L. reuteri strains. This was done by transforming pTRKH3-based vectors powering EGFP with 3 different promoters (ErmB, ldhL, slp) into Lactococcus lactis spp. Cremoris MG1363, Limosilactobacillus reuteri DSM 20016, and 5 different L. reuteri strains from chicken crops. The results showed that significant expression of EGFP can be achieved by the pTRKH3 vector in various strains, and especially in L. reuteri DSM20016, where regardless of the used promoter it resulted in good protein expression.

Figure 4. Vector selection.

With further research it was found that the pTRKH3 vector contains both an origin of replication (Ori) that is functional in E. coli, as well as one that allows for replication in L. reuteri. Though the latter at lower copy numbers, this will limit the metabolic load nanobody production would have on the organism, likely improving survivability in lung mucous. The identification of the nature of the ori in this plasmid was difficult, but through the use of nBLAST we determined it to be based on the pAMβ1 replication mechanism [6].

Additionally, since the vector contains a strongly expressed GFP gene, replacing this with whatever insert is desired will allow for easy recognition of false positives, since successful replacement changes colony color from fluorescent green to colorless. Therefore, this vector appears ideal for expression of the nanobody, and was chosen as the basis of our engineering.

Though this expression vector will be used for the final construct in L. reuteri, the pYTK001- and pYTK095 vectors from the Yeast Toolkit for Modular Assembly by Lee et al. [7], will be used as control and cloning backbones . These plasmids were designed to carry standardized Golden Gate Assembly constructs and contain a chloramphenicol and ampicillin resistance gene respectively, a p15 ori, and a superfolder GFP gene. This plasmid is far smaller (2676 bp) that the pTRKH3 expression vector (8961 bp) and has a very high copy number, thus allowing for more efficient cloning in E. coli and expression, on top of easy recognition of false positives when GFP is expressed.

Nanobody expression and secretion cassette

For the nanobody cassette itself several components are of importance. First a promoter, then the nanobody itself with fused tags, and finally a terminator. With this basic set-up both nanobody expression plasmids are designed.

First for the promoter, Lizier et al. [5], already compared three promising candidates, though these may be too strong for our purposes. It was therefore decided to use the weak constitutive Lacl promoter, as well as the IPTG inducible T7 promoter, the former being intended for final product use, and the latter being useful for expression and secretion testing.

Nanobody selection

For the final product the organism must express and secrete anti-hemagglutinin nanobodies. A number of candidates for our project revealed themselves in a study by Hufton et al. [20], where they generated and characterized 18 single-domain VHH nanobodies against the 2009 A(H1N1) avian influenza virus hemagglutinin through the use of an immune alpaca phage display library. A number of these nanobodies were binding but non-neutralizing, while others only neutralized H1N1, but most interestingly were the 5 variants with broad cross-subtype neutralizing capacity. Amongst these, variants R1a-B6 and R1a-A5 had consistently strong binding and neutralization in various avian influenza subtypes (see figure 5), and were therefore chosen as proof-of-concept nanobodies in our nanobody expression and secretion platform.

Figure 5. Binding of nanobody variants.

Additionally, the study experimented with turning R1a-B6 and R1a-A5 bivalent, which both increased neutralization efficacy to avian influenza variant A(H9N2) and gained the ability to neutralize variant A(H2N2). Therefore we decided to also test both of these nanobodies in parallel, using the (G5S)6 linker fused in between the two nanobodies and generating the variants R1a-B6-linker-B6, R1a-A5-linker-A4, and R1a-B6-linker-A5.

To confirm the ability of L. reuteri to produce functional nanobodies with our construct we also designed a control which produces anti-GFP nanobodies. This protein, which was originally developed by Rothbauer et al. [8], was selected based on a study by Kubala et al. [9], where a structural and thermodynamic analysis of the GFP:anti-GFP-nanobody complex was performed through x-ray crystallography and isothermal titration calorimetry respectively, showing tight binding of the nanobody to the fluorescent protein. With such well known binding, this anti-GFP nanobody is a perfect candidate as a control of nanobody expression and binding.

Figure 6. Interaction analysis of the GFP;GFP-nanobody complex.

This protein was expressed using the E. coli strain BL21-GOLD by Kubala et al. [9], we therefore codon optimized the gene for L. reuteri DSM20016. This was done using the OPTIMIZER tool created by Puigbò et al.[10], using the L. reuteri codon usage tablefrom the Kazusa codon usage database [11]. An added advantage of this is that we could express with less efficiency the codon optimised reuteri genes in E.coli.

Nanobody expression cassette design

Beyond the nanobody coding regions themselves, several additional aspects were considered during the design of the nanobody expression cassette, including the promoter, terminator and fused tags.

First, the promoter needed to be inducible to most effectively show whether nanobody expression has occurred. As one of the most commonly used promoters, the IPTG-inducible lac promoter was chosen for this purpose, for its well characterized nature, its relatively strong activity, as well as its common, effective and cheap induction chemical [12].

Next, besides nanobody expression, one of the goals of our project is the secretion of these peptides into extracellular space. For this purpose L. reuteri recognizes secretion signals which help guide proteins out of the cell. Therefore we chose the Cnb (Collagen-binding protein) secretion signal peptide for fusion to the N-terminal end of the nanobody, based on the comparative analysis of various L. reuteri secretion signals by Liu et al. [13]. In this study the efficacy of 29 secretion signals was analyzed by fusing them to lactonohydrolase from Rhinocladiella mackenziei, which is capable of hydrolyzing the mycotoxin ZEN (zearalenone). In this study, the Cnb secretion signal, here named S9, was found to be the most effective.

Figure 7. The signal peptide screening for the secretion expression system in L. reuteri.

Besides the secretion signal, it was also decided to fuse a 6x His-tag to the C-terminal end of the nanobodies to allow for more effective protein purification, which will become vital during the expression and secretion analysis. And for the anti-hemagglutinin constructs a Myc tag is also fused to the C-terminal end of the nanobody, which allows for an ELISA to be performed to confirm nanobody binding efficiency of our final products. Antibodies against this tag have already been developed long ago, for example by Evan et al. [14].

Lastly a terminator is required to end transcription. Though a number of efficient terminators are available, due to its ubiquitous and reliable nature it was decided to use the double terminator from biobrick BBa_B0015. This terminator efficiently ends transcription from both directions, which will become important in other constructs.

The end product of these assemblies are shown below in figure 8, where the arrangement of the expression cassettes are shown.

Figure 8. Arrangement of the expression cassettes.

Cloning

As mentioned above, various components are to be combined to produce the final nanobody expression and secretion plasmid. These components have been generated in 2 ways, PCR of existing plasmids or biobricks, and de novo synthesis. In this chapter the generation of these components will be described, as well as the assembly thereof into composite parts and plasmids.

Entry vector backbone

The entry vector used for the nanobody expression and secretion section of Nanobuddy was pYTK001, where PCR with p14 and p15 excised the superfolded GFP, amplifying the pMB1 ori and CmR, generating BBa_K4233000 as seen below in figure 9.

Figure 9. Creation of the Entry vector backbone for nanobody production and secretion BBa_K4233000.

Expression vector backbone

The expression vector used here was pTRKH3, where PCR with p5 and p6 excised mGFP5 and TetR, amplifying EryR, the pAMβ1 ori, and the p15 ori, generating BBa_K4233001 as seen below in figure 10.

Figure 10. Creation of the expression vector for nanobody production and secretion BBa_K4233001.

Antibiotic resistance marker

For selection of successful assemblies, an AmpR gene was also used. This fragment was synthesized de novo based on the AmpR from pYTK095. No PCR was used here, as in this stage of experimentation we did not have access to the pYTK095 plasmid, only the sequence. This synthesis generated BBa_K4233002, as seen below in figure 11.

Figure 11. Creation of the antibiotic resistance marker AmpR BBa_K4233002.

Anti-GFP nanobody expression cassette in the entry vector

For the assembly of the initial anti-GFP nanobody expression cassette entry vector construct, the BBa_K4233000 entry vector fragment, de novo synthesized BBa_K4233003 anti-GFP nanobody cassette, and BBa_K4233002 ampicillin resistance marker were Golden Gate Assembled as outlined before. This generated pRUC2001 (BBa_K4233008), as seen below in figure 12.

Figure 12. Assembly of anti-GFP nanobody expression cassette in the entry vector BBa_K4233008.

Anti-GFP nanobody expression cassette in the expression vector

For the assembly of the initial anti-GFP nanobody expression cassette expression construct, the BBa_K4233001 expression vector fragment, de novo synthesized BBa_K4233003 anti-GFP nanobody cassette, and BBa_K4233002 ampicillin resistance marker were Golden Gate Assembled as outlined before. This generated pRUC2006 (BBa_K4233013).

In addition, after expression-levels of the nanobodies with the lac promoter was deemed suboptimal, promoter exchange was attempted on BBa_K4233013 by performing PCR with p5 and p65, resulting in intermediate part BBa_K4233082, then circularizing this fragment with Golden Gate Assembly. This would have generated pRUC2206-T7 (BBa_K4233014), but cloning of this construct could not be performed in time (figure 13).

Figure 13. Assembly of anti-GFP nanobody expression cassette in the expression vector BBa_K4233014.

Anti-hemagglutinin nanobody expression cassette in the entry vector

For the assembly of the initial anti-hemagglutinin nanobody expression cassette entry vector constructs, the BBa_K4233000 entry vector fragment, de novo synthesized BBa_K4233004/BBa_K4233005/BBa_K4233006/BBa_K4233007/BBa_K42330077 anti-hemagglutinin nanobody cassettes, and BBa_K4233002 ampicillin resistance marker were Golden Gate Assembled as outlined before. This generated the pRUC2002 (BBa_K4233009), pRUC2003 (BBa_K4233010), pRUC2004 (BBa_K4233011), pRUC2005 (BBa_K4233012) and pRUC2036 (BBa_K42330078) plasmids. Of these constructs the BBa_K4233077 and BBa_K4233078 constructs were conceptual due to synthesis issues (figure 14).

Figure 14. Assembly of anti-hemagglutinin nanobody expression cassette in the entry vector BBa_K4233009 till BBa_K4233012 and BBa_K4233078.

Anti-hemagglutinin nanobody expression cassette in the expression vector

For the assembly of the initial anti-hemagglutinin nanobody expression cassette expression vector constructs, the BBa_K4233001 expression vector fragment, de novo synthesized BBa_K4233004/BBa_K4233005/BBa_K4233006/BBa_K4233007/BBa_K42330077 anti-hemagglutinin nanobody cassettes, and BBa_K4233002 ampicillin resistance marker were Golden Gate Assembled as outlined before. This generated the pRUC2007 (BBa_K4233015), pRUC2008 (BBa_K4233016), pRUC2009 (BBa_K4233017), pRUC2010 (BBa_K4233018) and pRUC2037 (BBa_K4233079) plasmids. Again, since BBa_K4233077 could not be synthesized, BBa_K4233079 could not be constructed.

Here too T7 promoter exchange was performed on all constructs using PCR with p5 and p65, generating the linear fragments BBa_K4233083, BBa_K4233084, BBa_K4233085, BBa_K4233086 and BBa_K4233087 (conceptual). Though the Golden Gate Assembly of these fragments into the pRUC2007-T7 (BBa_K4233019), pRUC2008-T7 (BBa_K4233020), pRUC2009-T7 (BBa_K4233021) and pRUC2010-T7 (BBa_K4233022) plasmids was attempted, a lack of time prevented its completion. Plasmid pRUC2037-T7 (BBa_K42330080) is again, conceptual due to the lack of BBa_K4233079 (figure 15).

Figure 15. Assembly of anti-hemagglutinin nanobody expression cassette in the expression vector BBa_K4233015 till BBa_K4233022, BBa_K4233079, and BBa_K4233080.

Testing

Assembly confirmation

Once these constructs had been assembled they were transformed into E. coli DH5ɑ and plated on LB plates with the relevant antibiotics. The resulting colonies were tested with colony PCR and brief exposure with UV light to determine assembly success (as false positives result in green fluorescence from the superfolded GFP of the uncut vectors), and those that pass this round were be inoculated in LB medium and the plasmids harvested for sequencing. When a successfully assembled plasmid had been found it was transformed, through electroporation, into L. lactis DSM20016 and E. coli BL21, after which expression and secretion was tested.

Nanobody secretion testing

Secretion testing will occur in L.reuteri since by design the secretion signal is codon optimized for our specific strain (L.reuteri DSM20016). However creation of the specific constructs will be performed in E.coli DH5α and E.coli BL21, since E.coli is much easier to cultivate than L.reuteri. According to literature L.reuteri should be a facultative anaerobic bacteria, but for easy and fast growth we chose for the molecular workhorse E.coli. The secretion testing itself is based on two primary research questions. Can the expression vector effectively use the nanobody cassette to express the nanobody in vivo? And does the secretion signal work effectively in the host bacteria?

To elucidate this, the L. lactis DSM20016 containing the expression and entry vectors with the nanobody cassettes will be used for protein expression, and the resulting cultures are then pelleted through centrifugation. The supernatant and pellet will be collected separately and the latter is disrupted through sonification to maintain protein structure integrity. The protein suspensions are centrifuged again to remove any remaining cell debris and used for SDS-PAGE and western blot to determine the presence of the nanobodies inside and outside the cells. This will both show in principle the viability of expression of the nanobodies and the subsequent secretion thereof.

For testing the strength of the secretion signal we would collect again protein suspensions from the supernatant and the cell lysis mixture. However instead of performing SDS-PAGE or western blot, we would firstly perform protein purification with a Nickel beads column and a Bradford assay with a BSA control. From this we should be able to determine purified protein concentrations from both inside and outside the cells, and thus do a fractional analysis to conclude on the strength of the secretion signal.

Nanobody efficacy testing

When expression and secretion is determined to be successful, the nanobody binding efficiency can be quantified using direct ELISA (see figure 16 below) [15, 16]. The nanobodies contain MIC tags and his-tags, which both have known antibodies with bound reporter enzymes in this case Horseradish peroxide (HRP). By coating the 96 wells maxisorp plates with GFP and hemagglutinin proteins and incubating them with known concentrations of the anti-GFP and anti-hemagglutinin nanobodies, the binding is facilitated. Then by subsequent incubation of the secondary antibodies which are already bound to the reporter enzymes and the addition of a chemiluminescent substrate after washing. We are able to determine the binding affinity of previously mentioned antibodies respectively based on the fluorescence of the converted chemiluminescent substrate.

Figure 16. Direct ELISA to quantify the nanobody binding efficiency

Biocontainment

Though the primary goal of this project was to genetically modify L. reuteri to express and secrete anti-influenza nanobodies, its in situ application will both involve spraying of the organism into the environment, and its subsequent survival exclusively in the lungs of chickens. Since it is possible for bacteria to incorporate foreign DNA into its own genome, it would be possible for the DNA of a genetically engineered organism to enter the environment and be absorbed and utilized by other species in unexpected ways. This is unlikely to have harmful effects, but since this is not an absolute certainty, it is wise to limit the release of modified organisms and their DNA into nature, which is no different with this project.

Design considerations

So to ensure that the organism only survives inside chicken lungs a killswitch mechanism was designed using two environmental factors, high temperature and the absence of light. After all, chicken lungs are both warm (41℃) and dark, which are unusual environmental factors not often found outside living organisms, especially in the Netherlands. So when these mechanisms are employed, only those two factors will allow the organism to not die.

Usually utilizing only one of these two is enough, however due to the non-binary activation behaviour of temperature and light switches, and the decreased likelyhood of escape when using to separate killswitches, both were designed in parallel but can also be used individually.

Killing mechanism

As mentioned above, the primary goal of the kill-switches is to kill the microorganism under specific circumstances. Naturally this requires a killing mechanism, of which a number of possible variants are available. One more well known type is a toxin/antitoxin mechanism, where a toxin is constitutively expressed while an antitoxin is activated by some external signal. Though this is a well documented method, it is not always the most effective. Therefore we decided it was interesting to try a CRISPR-Cas9-based killing mechanism, which could have the capacity for greater killing efficiency and greater adaptability.

This was explored in a study by Rottinghaus et al [17], where a constitutive Pcon promoter controls a tetR gene, which inhibits constitutive Ptet promoters controlling a gRNS and a Cas9 gene. TetR is inhibited by aTc (anhydrotetracycline), so adding this molecule to the organism containing the construct would result in the inability of TetR to inhibit the Ptet promoters, subsequently allowing gRNA and Cas9 to express and to form a CRISPR-Cas9 complex, killing the cell (figure 17).

Figure 17. Schematic of the CRISPR-Cas9 kill mechanism by Rottinghause et al [17].

Kill mechanism testing showed a significantly decreased colony forming unit count when aTc was added, though the effectiveness varied based on the specific target selected. Therefore it was chosen to have our own sgRNA target various conserved regions of L. reuteri DSM 20016 16S rRNA, and compare effectiveness, as this gene is highly conserved in bacterial species to begin with, allowing for the potential adaptation of the kill switches into species other than L. reuteri.

Similar to our intent to seed the organism in chicken lungs, the paper also tested in vivo killswitch effectiveness, by seeding the gut of mice with their modified E. coli Nissle 1917 strain. Feeding the mice with aTc resulted in the rapid decrease in modified E. coli CFUs recovered from feces, as can be seen in Figure 18 below.

Figure 18. Effectiveness of the kill mechanism in vivo.

Light killswitch

In order to utilize our chosen kill mechanism we need a way to convert environmental stimuli into activation of expression of the cas9, while the sgRNA is constantly transcribed. As said before one of the chosen environmental factors is light, however because of the nature of photons, light is difficult to detect and often requires cofactors not usually synthesized by bacteria. To engineer a bacterium to allow for the synthesis of these cofactors frequently requires many different enzymes and thus cloning steps which would not be achievable within this project. This is why we use the natural ability of L. reuteri to produce cobalamin (Vitamin B12), which can detect light and together with the transcription factor CarH has the ability to regulate gene expression. CarH is a bacterial transcription factor used by Myxococcus xanthus and Thermus thermophilus, which binds Adenosylcobalamin (a form of B12 produced by L. reuteri) and forms a tetramer allowing the complex to bind DNA and thus stop transcription (figure 19). Light triggers the dissociation of the tetramer and thus allows gene expression.

Figure 19. Schematic visualization of the dissociation of the CarH tetramer triggered by light.

In order to combine the light sensor and kill mechanism into a light activated killswitch we designed the cassette as in figure 20. Here we placed the constitutive strong promoter J23102 to produce the TtCarH repressor derived from Thermus thermophilus, which in turn represses pCarH promoter activity localized upstream of Cas9. When exposed to light the CarH will dissociate and release the promoter allowing transcription of the Cas9 which will bind the sgRNA constitutively transcribed by J23100 and start cutting the 16S rRNA which will kill the cell. J23100 is used to control constitutive sgRNA transcription, because in nature it controls shRNA (small hairpin RNA) transcription, and therefore does not add the poly-A tail to the RNA which would normally mediate degradation of the molecule. This also positions the sgRNA at the end of the cassette allowing for easy substitution when this cassette is utilized in a different organism.

Figure 20. scheme of the light killswitch cassette BBa_K4233069.

Temperature killswitch

The other variable we looked at to distinguish between the environment of the chicken lung and the outside world is temperature. The body temperature of a chicken is between 40 and 42 degrees celsius whereas the outside temperature, especially in the Netherlands, will rarely reach this high. We decided upon the TlpA system derived from Salmonella typhimurium. This protein forms a dimer which binds its associated promoter and stops transcription, however at high temperatures the dimer becomes unstable releasing the promoter and allowing gene expression. The main upside of this system is that the transition temperature is around 41 degrees celsius which is similar to the body temperature of the chicken but not a human, and results in a 200-fold increase in expression. Furthermore this system has previously been used to create a kill switch in E. coli and has already been engineered to transition at 39 and 36 degrees celsius by Piraner et al [18], which could prove useful in fine tuning the system. To further tune the system we test 2 different promoters, one is the native promoter of TlpA and the other an adapted version. This is because the native promoter includes a peculiar region between the -35 and -10 boxes consisting of only 2 nucleotides per strand, guanine and thymine on the one side and cytosine and adenine on the other. The reason for this is unclear but it is hypothesized that it aids the binding of TlpA dimer. The adapted version of the promoter changes this region for the same region of the pbu promoter, a promoter native to L. reuteri and involved in the genes responsible for glycerol metabolism. We hypothesize that this change will decrease binding of the protein and thus cause more leakage but also provide a higher uninhibited expression of the gene, this will then serve as a proof-of-concept that these parameters can be fine tuned to create a more effective and safe use of our treatment.

Figure 21. Schematic visualization of the regulation of pTlpA at increasing temperature.

As can be seen in the figure 21 above this system is deactivated at low temperature and activates as the temperature increases. If we were to use the same cassette setup as with the CarH system we would get the opposite result to what we want to achieve, the bacteria will survive at low temperatures while it dies as it enters the lung. This is why we need to invert the signal, which can be done using TetR. TetR is a protein that represses the ptet promoter, and thus can invert a signal resulting in the activation of the kill mechanism at low temperatures.

We designed the temperature killswitch cassette based on the design by Piraner et al [18], as this has been proven to work. Accordingly the 3 TlpA variants of TlpA36, TlpA39, and TlpA41 are coexpressed with TetR by the 2 variants of the pTlpA promoter mentioned before. This creates 6 variants of the killswitch showing the adaptability of the system. The TetR is expressed at high temperatures and proceeds to inhibit the Tet promoter, but at low temperatures the TetR is repressed allowing for expression of the Cas9. The Cas9 then binds the constitutively expressed sgRNA and proceeds to kill the cell (figure 22).

Figure 22. Scheme of the temperature killswitch cassettes BBa_K4233063 till BBa_K4233068.

Cloning strategy

As mentioned above, various components are to be combined to produce the final killswitch plasmids. These components have been generated in 2 ways, PCR of existing plasmids or biobricks, and de novo synthesis. In this chapter the generation of these components will be described, as well as the assembly thereof into composite parts and plasmids

Entry vector backbone

The Entry Vector used in creating the kill switch for Nanobuddy was pYTK095. The superfolded GFP has been excised by PCR using p48 and p50 to generate BBa_K4233025, p49 and p50 to generate BBa_K4233026, or p50 and p62 to generate BBa_K4233027. All backbones include the AmpR and pMB1 ori, while only the later two also include the BBa_B1005 terminator. The difference between BBa_K4233026 and BBa_K4233027 is the BsaI cut-site used during Golden Gate Assembly, as can be seen below in figure 23.

Figure 23. Creation of kill switch Entry vector backbone BBa_K4233025 till BBa_K42330.

Expression vector backbone

The expression vector was created from pTRKH3-ermGFP, using p5 and p65 to excise the GFP, TetR and EryR and amplify the pAMβ1 ori and p15 ori. This generates part BBa_K4233029, as can be seen below in figure 24.

Figure 24. Creation of kill switch expression vector backbone BBa_K4233029.

Antibiotic resistance marker

The antibiotic resistance marker used to create the killswitch is PCRed from pYTK001 using p51 and p52, this generates part BBa_K4233028 as can be seen below in figure 25.

Figure 25. Creation of antibiotic resistance marker fragment creation BBa_K4233028.

mRPF1

The mRFP1 fragments used to test the promoters of the killswitch were created using two different methods. The biobrick BBa_K516030 was PCRed using p30 and p31 to create a fragment containing the mRFP1 and a double terminator (BBa_K4233023). While the other fragment containing the mRFP1 and the BBa_B1002 terminator was synthesized de novo generating part BBa_K4233024 (figure 26).

Figure 26. Creation of mRFP1 fragments BBa_K4233023 and BBa_K4233024.

Test Plasmids

We now know out of which promoters our killswitch will be built, but before this system can be assembled its promoters first need to be tested. We did this by making them express mRFP1 and measuring the fluorescence. This gives us insight into the strength and transition values of the promoters.

Constitutive promoter test

For the assembly of both constitutive promoter test plasmids the promoters BBa_J23100 and BBa_J23102 were de novo synthesized to generate parts BBa_K4233030 and BBa_K4233031 respectively. These were combined with the BBa_K4233023 mRFP1 fragment and the BBa_K4233025 backbone to create pRUC2211 (BBa_K4233041) and pRUC2212 (BBa_K4233042), as shown below in figure 27.

Figure 27. Assembly of the constitutive promoter test plasmids BBa_K4233041 and BBa_K4233042.

TetR test

For the assembly of the TetR test plasmid a fragment containing the J23102 promoter, the TetR gene, a double terminator, and the Tet promoter was synthesized to generate part BBa_K4233033. This was combined with the BBa_K4233024 mRFP1 fragment and the BBa_K4233026 backbone to create pRUC2214 (BBa_K4233044), as shown below in figure 28.

Figure 28. Assembly of the TetR test plasmid BBa_K4233044.

TlpA test

For the assembly of the TlpA test plasmids the fragments containing all 6 combinations of the 2 promoters and the 3 TlpA variants were synthesized to generate parts BBa_K4233034, BBa_K4233035, BBa_K4233036, BBa_K4233037, BBa_K4233038, and BBa_K4233039. These were combined with the BBa_K4233023 mRFP1 fragment and the BBa_K4233025 backbone to create pRUC2215 (BBa_K4233045), pRUC2216 (BBa_K4233046), pRUC2217 (BBa_K4233047), pRUC2218 (BBa_K4233048), pRUC2219 (BBa_K4233049), and pRUC22120 (BBa_K4233050) respectively, as shown below in figure 29.

Figure 29. Assembly of the TlpA test plasmids BBa_K4233045 till BBa_K4233050.

CarH promoter test

For the assembly of the CarH promoter test plasmid a fragment containing the pCarH promoter should have been synthesized generating part BBa_K4233032. However it could not be synthesized and could not be optimized due to time constraints, but for further inquiries we recommend contacting the writers of this page. If obtained, the part would have been combined with the BBa_K4233023 mRFP1 fragment and the BBa_K4233025 backbone to create pRUC2213 (BBa_K4233043) , as shown below in figure 30.

Figure 30. Assembly of the CarH promoter test plasmid BBa_K4233043.

CarH test

For the assembly of the CarH test plasmid a fragment containing the J23102 promoter, the CarH gene, and the pCarH promoter should have been synthesized generating part BBa_K4233040. However as said before this was not possible within the given time frame. If obtained, the part would have been combined with the BBa_K4233024 mRFP1 fragment and the BBa_K4233026 backbone to create pRUC2221 (BBa_K4233051), as shown below in figure 31.

Figure 31. Assembly of the CarH test plasmid BBa_K4233051.

Final temperature kill-switch

The final kill-switches are assembled in 3 steps: The kill-switch cassette assembly, the kill-switch cassette amplification, and then kill-switch expression vector incorporation.

Temperature kill-switch cassette assembly

To assemble the final temperature killswitch cassette, several components are needed. First are the 5 temperature sensor fragments, pTlpA1-TlpA36C (BBa_K4233034), pTlpA1-TlpA39C (BBa_K4233035), pTlpA1-TlpA41C (BBa_K4233036), pTlpA2-TlpA36C (BBa_K4233037), pTlpA2-TlpA39C (BBa_K4233038) and pTlpA2-TlpA41C (BBa_K4233034). Next, TetR inverter component BBa_K4233033 is PCRed using p34 and p35 to remove the constitutive J23102 promoter. Biobrick BBa_2457006 (from iGEM17_Amazonas_Brazil) is PCRed using p55 and p57 to amplify only the Cas9 coding region, and to add a downstream BBa_B1002 terminator, generating BBa_K4233053. Lastly the sgRNA cassette targeting L. reuteri 16S rRNA has been synthesized de novo. All these components are used in a Golden Gate Assembly with entry vector fragment BBa_4233027 to form pRUC2222 (BBa_K4233056), pRUC2223 (BBa_K4233057), pRUC2224 (BBa_K4233058), pRUC2225 (BBa_K4233059), pRUC2226 (BBa_K4233060) and pRUC2227 (BBa_K4233061) respectively. See figure 32 below.

Figure 32. Assembly of temperature kill-switch cassette BBa_K4233056 till BBa_K4233061.

Temperature kill-switch cassette amplification

Once all the individual components have been assembled into pRUC2222-pRUC2227, these plasmids can be used as a template to amplify the entire killswitch cassette. This region, from the temperature sensor component to the sgRNA cassette are to be PCRed using p36 and p59, generating the killswitch cassette fragments pTlpA1-TlpA36C (BBa_K4233063), pTlpA1-TlpA39C (BBa_K4233064), pTlpA1-TlpA41C (BBa_K4233065), pTlpA2-TlpA36C (BBa_K4233066), pTlpA2-TlpA39C (BBa_K4233067) and pTlpA2-TlpA41C (BBa_K4233068). See figure 33 below.

Figure 33. Amplification of Temperature kill-switch cassette BBa_K4233063 till BBa_K4233068.

Temperature kill-switch expression vector incorporation

Finally, with the temperature killswitch cassettes amplified, they can be incorporated into the expression vector. This is done by performing Golden Gate Assembly on the cassettes pTlpA1-TlpA36C (BBa_K4233063)/pTlpA1-TlpA39C (BBa_K4233064)/pTlpA1-TlpA41C (BBa_K4233065)/pTlpA2-TlpA36C (BBa_K4233066)/pTlpA2-TlpA39C (BBa_K4233067)/pTlpA2-TlpA41C (BBa_K4233068), the chloramphenicol resistance marker BBa_K4233028, and the expression vector BBa_K4233029. The resulting constructs are pRUC2229 (BBa_K4233070), pRUC2230 (BBa_K4233071), pRUC2231 (BBa_K4233072), pRUC2232 (BBa_K4233073), pRUC2233 (BBa_K4233074) and pRUC2234 (BBa_K4233075) respectively, as can be seen below in figure 34. These are the final kills-switch constructs ready for testing.

Figure 34. Assembly of temperature kill-switch expression vector BBa_K4233070 till BBa_K4233075.

Final light kill-switch

Here too, the final kill-switches are assembled in 3 steps: The kill-switch cassette assembly, the kill-switch cassette amplification, and then kill-switch expression vector incorporation. The assembly of the light kill-switch is highly similar to the temperature kill-switches, and differs only in the sensor component and the overhangs used for the Cas9 component. It will therefore be described in a more brief fashion.

Light kill-switch cassette assembly

The light kill-switch cassette was assembled using the de novo synthesized CarH light sensor component (BBa_K4233040), the Cas9 coding region amplified using p56 and p57, the sgRNA cassette (BBa_K4233055) and the entry vector (BBa_K4233027), to form the plasmid pRUC2228 (BBa_K4233062) as shown below in figure 35.

Figure 35. Assembly of light kill-switch cassette BBa_K4233062.

Light kill-switch cassette amplification

The light kill-switch cassette is then amplified by performing PCR on pRUC2228 using p34 and p59, generating fragment BBa_K4233069 as shown below in figure 36.

Figure 36. Amplification of light kill-switch cassette BBa_K4233069.

Light kill-switch expression vector incorporation

Finally, with the light killswitch cassette amplified, it can be assembled into the final light kill-switch through Golden Gate Assembly with the chloramphenicol resistance marker BBa_K4233028 and the expression vector BBa_K4233029, to form pRUC2225 (BBa_K4233076) as can be seen below in figure 37. This is the final light kill-switch construct ready for testing.

Figure 37. Assembly of light kill-switch expression vector BBa_K4233076.

Testing

Assembly confirmation

Once these constructs had been assembled they were transformed into E. coli DH5ɑ and plated on LB plates with the relevant antibiotics. The resulting colonies were tested with colony PCR and brief exposure with UV light to determine assembly success, as false positives result in green fluorescence from the superfolded GFP of the uncut vectors, while successful assemblies could exhibit red fluorescence from the mRFP1. Those that pass this round were inoculated in LB medium and the plasmids harvested for sequencing. When a successfully assembled plasmid had been found it was transformed, through electroporation, into L. lactis DSM20016 and E. coli BL21, after which construct functionality could be tested.

Primary construct testing

The first round of assemblies involves the insertion of the individual promoters, the TetR inverter, the temperature sensors and the light sensor into pYTK095-based mRFP1 quantification constructs. With these constructs the individual components of the final kill switches could be quantified. So after assembling these plasmids we measured for red fluorescence at various temperatures.

To do this the bacteria containing the constructs were first grown overnight in LB medium with ampicillin, then diluted 1:200 into new LB medium with ampicillin to achieve exponential growth. These cultures were diluted to OD600 0.4-0.6±0.3 in LB- or M9 medium with ampicillin and pipetted in triplicate into a 96-well microtiter plate. These plates were incubated in temperature-controlled microtiter plate readers at various temperatures. Though initially the desire was to test at 26℃, 30℃, 35℃, 37℃, 39℃, 41℃ and 43℃, only 4 experiments could be performed in time: 26℃, 30℃, 37℃ and 41℃. See Results for the data of these experiments. Had the light sensor been available, a similar experiment would have been developed where the cells would be exposed to different intensities of light at different wave-lengths, to determine sensor sensitivity.

Due to the time constraints and synthesis complications the final killswitch constructs could not be assembled, but had they been constructed a series of growth-curves would have been made at different temperatures, showing whether the temperature kill-switches would inhibit cell-growth, and to what extent. Based on this data a preliminary analysis of kill-switch efficacy could have been performed, which could then have been optimized using promoter exchange, vector variation, and the modification of other conditions.

For the light kill-switch the same could have been done, but using light-exposure instead of variable temperature. Different light intensities and wave-lengths could have been tested to characterize kill-switch sensitivity.

Temporary deactivation mechanism

Since the killswitch induces cell-death when exposed to light and low temperature, the spray delivery mechanism would likely result in the death of the cells before delivery could be completed. Hence the need for a temporary deactivation mechanism is required. This mechanism is based on the lac repression system, where the lac repressor is capable of binding to a lac repressor binding site on the DNA, preventing transcription from progressing. Cloning this lac repressor between the promoter and the Cas9 of the killswitch would result in the prevention of Cas9 expression when the lac repressor is expressed. Simply controlling lac repressor expression with an inducible promoter like T7 would then allow for tight killswitch control. The inducer simply needs to be added to the liquid containing Nanobuddy, which results in killswitch deactivation, after which the product is sprayed over chickens. Then once the droplets reach a surface the inducer will be diluted over time causing the deactivation mechanism to be turned off.

Deactivation could quite easily be tested with IPTG if the T7 promoter were to be used, however, IPTG cytotoxicity would need to be tested, as well as system leakiness. Due to time constraints this concept is therefore left conceptual for future research. In this research the following constructs are to be developed:

Figure 38. Temporary deactivation mechanism.

First, IPTG deactivation of the Lacl repressor is to be tested, requiring knowledge of Placl activity. Hence a Placl-controlled GFP is to be constructed (A). Next, the repressor itself can be tested, where the previously used J23100 promoter (strong) controls Lacl repressor expression, which in turn represses GFP expression. Finally, the deactivation mechanism itself can be developed (B), where the Lacl operator is cloned between the Ptet or PcarH promoter and the Cas9 gene. The inducible promoter and Lacl repressor gene itself is simply placed anywhere on the plasmid.

Genomic integration

Though plasmids are widely used as carriers for various modified genes and constructs, these require selection pressure to remain functional, often applied using antibiotics and respective antibiotic resistance markers. However, inside the lungs of chickens the application of such antibiotics is unfeasible and unsafe. Hence another method must be used to ensure the retention of desirable genetic material. Genomic integration is the easiest method for which various tools are available. CRISPR-Cas9 combined with HDR (homology directed repair) is the simplest method, however, this requires the in-vivo expression of Cas9 to function, which would activate the killswitch mechanism resulting in cell-death. Therefore, it is recommended to instead use Lambda red recombinase-mediated genomic integration instead, which involves the expression of Lambda red recombinase in the presence of dsDNA with homologous end-regions identical to the recombination target sequence, resulting in site-directed recombination and gene insertion. This method allows for the use of relatively high molecular weight integration cassettes of at least 50 kb, which is ideal for our large constructs [19].

Engineering cycle

As can be seen in more detail in the Results page, the nanobody expression constructs were tested on expression capacity, but very little of our desired protein was found on the SDS-PAGE gels. We attempted to optimize the expression in various ways, but this did not help much. So in the end we decided to overexpress our nanobodies by replacing the initial IPTG inducible Lac promoter with the more powerful T7 promoter.

To do this we designed only a single primer (p65), targeting the RBS and beginning of the nanobody coding gene, and possessing a long non-binding tail containing the T7 promoter and a BsaI restriction site. As a reverse primer the existing p5 primer was used. When PCR is performed with these primers on pRUC2206-10 and 37, the Lac promoter is excised, and the BsaI restriction site and T7 promoter are added in front of the nanobody gene instead, requiring only Golden Gate Assembly with this one DNA fragment to re-circularize and have the promoter be fully exchanged, forming pRUC22(06-10)-7 and pRUC2237-T7. See figure 39 below for the binding of primer p65, figure 40 for an example of the PCR on pRUC2206, and figure 41 for the overview of the used plasmids and parts.

Figure 39. Binding of primer p65.

Figure 40. Example of the PCR on pRUC2206.

Figure 41. Overview of the used plasmids and parts.

In figure 41 the intermediate linearized fragments have also been marked as individual parts to allow for proper registration in the iGEM registry page.

Due to time constraints, despite attempting, we were not able to successfully assemble and transform the T7 replacement plasmid, however this represents an example of our engineering cycle, as we constructed a plasmid, tested it for activity, then used the collected information to improve the assembly.

Part improvement

While all the constructs we designed and built in our project were based on existing developed parts and research, one component in particular is of note. The sgRNA part BBa_K2457002 was designed by iGEM17_Amazonas_Brazil to target LacZ, and contains the BBa_J23100 promoter followed by the optimized sgRNA and the L3S2P21 terminator, all in the standard biobrick format. For our purpose we synthesized the sgRNA without the biobrick prefix and suffix, with a crRNA specific for L. reuteri 16S rRNA, and with a BsaI restriction site on either end of the fragment cutting inward, allowing for incorporation into our kill-switches (BBa_K4233055).

Though adaptation of the Brazilian biobrick to target a different sequence was not difficult, de novo synthesis of dsDNA can become quite expensive if a library of sgRNA is to be generated. To solve this problem we modified the biobrick BBa_K2457002 by replacing the 20 bp crRNA region with two outward-facing BbsI type-IIS restriction sites and 2 bp of spacers, resulting in the formation of a self-excision module (see figure 42). All that is needed to insert a library is to clone this part into a plasmid of choice, then order two primers containing the crRNA library and overhangs, annealing these primers to form dsDNA with overhangs, and performing a Golden Gate Assembly with the annealed crRNA library and the BBa_K4233081 part (figure 43).

Figure 42. A self-excision module.

Figure 43. Insertion of a library.

Due to time constraints this construct has not been cloned, but it has been designed in-silico and uploaded to the registry, allowing for easy synthesis by future iGEM teams.

References

[1] Ngunjiri, J.M. et al. (2019) “Farm stage, bird age, and body site dominantly affect the quantity, taxonomic composition, and dynamics of respiratory and gut microbiota of commercial layer chickens,” Applied and Environmental Microbiology, 85(9). Available at: https://doi.org/10.1128/aem.03137-18.

[2] Krüger, C. et al. (2002) “In situ delivery of passive immunity by lactobacilli producing single-chain antibodies,” Nature Biotechnology, 20(7), pp. 702–706. Available at: https://doi.org/10.1038/nbt0702-702.

[3] J5 manual - JBEI (no date) The Golden Gate assembly method (and MoClo and GoldenBraid). Available at: https://j5.jbei.org/j5manual/ (Accessed: September 24, 2022).

[4] Gene synthesis: IDT (no date) Integrated DNA Technologies. Available at: https://www.idtdna.com/pages/products/genes-and-gene-fragments/custom-gene-synthesis (Accessed: October 1, 2022).

[5] Lizier, M. et al. (2010) “Comparison of expression vectors in lactobacillus reuteri  strains,” FEMS Microbiology Letters, 308(1), pp. 8–15. Available at: https://doi.org/10.1111/j.1574-6968.2010.01978.x.

[6] Nucleotide blast: Search nucleotide databases using a nucleotide query (no date) National Center for Biotechnology Information. U.S. National Library of Medicine. Available at: https://blast.ncbi.nlm.nih.gov/Blast.cgi?PROGRAM=blastn&BLAST_SPEC=GeoBlast&PAGE_TYPE=BlastSearch.

[7] Lee, M.E. et al. (2015) “A highly characterized yeast toolkit for modular, Multipart Assembly,” ACS Synthetic Biology, 4(9), pp. 975–986. Available at: https://doi.org/10.1021/sb500366v.

[8] Rothbauer, U. et al. (2008) “A versatile Nanotrap for biochemical and functional studies with fluorescent fusion proteins,” Molecular & Cellular Proteomics, 7(2), pp. 282–289. Available at: https://doi.org/10.1074/mcp.m700342-mcp200.

[9] Kubala, M.H. et al. (2010) “Structural and thermodynamic analysis of the GFP:GFP-Nanobody Complex,” Protein Science, 19(12), pp. 2389–2401. Available at: https://doi.org/10.1002/pro.519.

[10] Puigbo, P. et al. (2007) “Optimizer: A web server for optimizing the codon usage of DNA sequences,” Nucleic Acids Research, 35(Web Server). Available at: https://doi.org/10.1093/nar/gkm219.

[11] Codon Usage Database (no date) Codon usage table. Available at: http://www.kazusa.or.jp/codon/cgi-bin/showcodon.cgi?species=1598&aa=1&style=N (Accessed: September 15, 2022).

[12] Hansen, L.H., Knudsen, S. and Sørensen, S.J. (1998) “The effect of the lacy gene on the induction of IPTG inducible promoters, studied in escherichia coli and pseudomonas fluorescens,” Current Microbiology, 36(6), pp. 341–347. Available at: https://doi.org/10.1007/s002849900320.

[13] Liu, F. et al. (2019) “Biodetoxification of fungal mycotoxins Zearalenone by engineered probiotic bacterium lactobacillus reuteri with surface-displayed lactonohydrolase,” Applied Microbiology and Biotechnology, 103(21-22), pp. 8813–8824. Available at: https://doi.org/10.1007/s00253-019-10153-1.

[14] Evan, G.I. et al. (1985) “Isolation of monoclonal antibodies specific for human c-myc proto-oncogene product.,” Molecular and Cellular Biology, 5(12), pp. 3610–3616. Available at: https://doi.org/10.1128/mcb.5.12.3610.

[15] What is a Direct ELISA? (2020) AAT Bioquest. Available at: https://www.aatbio.com/resources/faq-frequently-asked-questions/What-is-a-Direct-ELISA (Accessed: October 3, 2022).

[16] Lin, A.V. (2015) “Direct elisa,” Methods in Molecular Biology, pp. 61–67. Available at: https://doi.org/10.1007/978-1-4939-2742-5_6.

[17] Rottinghaus, A.G. et al. (2022) “Genetically stable CRISPR-based kill switches for engineered microbes,” Nature Communications, 13(1). Available at: https://doi.org/10.1038/s41467-022-28163-5.

[18] Piraner, D.I. et al. (2016) “Tunable thermal bioswitches for in vivo control of Microbial Therapeutics,” Nature Chemical Biology, 13(1), pp. 75–80. Available at: https://doi.org/10.1038/nchembio.2233.

[19] Juhas, M. and Ajioka, J.W. (2016) “Lambda red recombinase-mediated integration of the high molecular weight DNA into the escherichia coli chromosome,” Microbial Cell Factories, 15(1). Available at: https://doi.org/10.1186/s12934-016-0571-y.

[20] Hufton, S. E. er al. (2014) “The Breadth of Cross Sub-Type Neutralisation Activity of a Single Domain Antibody to Influenza Hemagglutinin Can Be Increased by Antibody Valency” PLoS One. 2014; 9(8): e103294. Available at: https://doi.org/10.1186/s12934-016-0571-y.