Team:ETH Zurich/core/results
Contents
- 1 Results
- 1.1 Results Overview
- 1.2 Engineering Bacteriophages in Yeast
- 1.2.1 Design Overview
- 1.2.2 Stable and Selectable YAC-T7 Vector
- 1.2.3 Formation of Functional T7 Phages from the YAC-T7 vector
- 1.2.4 Induction of a Double Stranded Break in the YAC-T7 Vector by CRISPR/Cas9
- 1.2.5 Swapping T7 Tail Fiber Region by its T3 Homologue
- 1.2.6 Making Phage Libraries with the Yeast Approach
- 1.2.7 Bibliography
- 1.3 Recombineering
- 1.3.1 Design Overview
- 1.3.2 Phage genome editing by homologous recombination
- 1.3.3 Host range alteration by tail fiber editing
- 1.3.4 Plasmid library generation
- 1.3.5 Phage library generation
- 1.3.6 Construction of E. coli resistant to T7 with a CRISPR-Cas system targeting the tail fiber
- 1.3.7 Library contains variants capable of infecting CRISPR-resistant bacteria
- 1.3.8 Selection of good variants with the bioreactor
- 1.4 In vitro
- 1.5 Phage training
Results
<nav>
</nav>
Results Overview
Genome Editing
Phage Formation
Infectivity Swap
We first aimed to verify that the T7 host spectrum can be modified by modifying its tail fiber. It has been shown that by changing the T7 tail fiber to the T3 homologue, the host range can be altered. The E. coli ECOR16 strain is not infected by T7 but is by T3. However, by exchanging the C-terminal domain of the tail fiber of the T7 phage to the T3 homologue, the resulting hybrid phage can infect [ref]. We successfully reproduced this change in host specificity (Fig. ).
<figure class="figure-center">
<img src="
">
<figcaption>Figure : Spot assays of T7, T3 and T7 engineered to carry the C-terminal tail fiber domain of T3 on E. coli DH5alpha and E. coli ECOR16. The host ranges of T7, T3 and T7 carrying the C-terminal domain of the T3 tail fiber were assessed by spotting them on plates overlaid with E. coli DH5alpha and E. coli ECOR16.</figcaption>
</figure>
We further attempted to find which amino acids were responsible for the specificity change. By aligning the tail fiber sequences of T7 and T3 we found that the C-terminal domain is less conserved, while the N-terminal domain is more conserved (Fig. ). Certain unconserved sequences in the C-terminal domain were selected and exchanged individually (Fig. ). The infectivities of these hybrid phages toward E. coli ECOR16 were measured. We found that only the hybrid with the entire C-terminal domain of T3 is able to infect ECOR16 (Fig. ).
<figure class="figure-center">
<img src="
">
<figcaption>Figure : Alignment of the T7 and T3 tailfiber proteins. The amino acid sequences of the T7 and T3 proteins were aligned. Full red bars denote conservation, while half red bars denote differences between the two proteins. With the blue bar we denote the C-terminal domain that was exchanged in T7 for the T3 variant to change the host specificity.</figcaption>
</figure>
<figure class="figure-center">
<img src="
">
<figcaption>Figure . Top: Alignment of the amino acid sequences C-terminal domains of the T7 and T3 tail fiber. Full red bars denote conservation, while half red bars denote differences between the two proteins. With blue bars we denote the unconserved sequences that were individually exchanged in T7 for the T3 sequence. Middle: The positions of the exchanged sequences in the 3D structure [ref] of the tail fiber protein. Bottom: Infectivities of the different hybrid phages on E. coli ECOR16, a strain that is infected by T3 but not by wild type T7, measured by performing plaque assays with recombinant phages generated by the recombineering approach.</figcaption>
</figure>
Library
Engineering Bacteriophages in Yeast
Design Overview
The excellence of Saccharomyces cerevisiae in repairing double stranded DNA breaks by homologous recombination allows for the efficient insertion of linear DNA fragments at any locus of interest in the genome or on centromeric plasmids. We aim to use this property to establish a system that efficiently randomizes the gp17 tail fiber protein of the bacteriophage T7. Gap repair in yeast has been successfully used to assemble modified T7 genomes with a yeast artificial chromosome (YAC), that allows for the propagation and selection of our assembled YAC-T7 vector. Electroporation into bacteria of the extracted YAC-T7 vector leads to the rebooting of the phage genome and the formation of functional phages [1]<a style="color: #ffffff; text-decoration:none;" href="#biblio-yeast">Ando, H., Lemire, S., Pires, D. P., & Lu, T. K. (2015). Engineering Modular Viral Scaffolds for Targeted Bacterial Population Editing. Cell systems, 1(3), 187–196. doi:10.1016/j.cels.2015.08.013</a>.
Inspired by this system, we first assembled a stable and selectable YAC-T7 vector. Then, CRISPR/Cas9 was used to induce a double stranded break in the T7 genome at the site of interest. By providing oligos with degenerated base pairs flanked by regions that are homologous to the cut site, S. cerevisiae will repair the double stranded break with the provided randomized DNA and rescue the selection marker. The resulting modified YAC-T7 vector was extracted from yeast and either electroporated into bacteria or prepared for use in a cell free transcription-translation system in order to generate the new phages containing different variants of the gp17 tail fiber protein (Figure 1).
<figure class="figure-center">
<img src="
" alt="yeast overview">
<figcaption> Figure 1: Overview of the Yeast Approach. A stable and selectable YAC-T7 vector and the CRISPR/Cas9 system are introduced into yeast. Induction of the CRISPR/Cas9 system by β-estradiol leads to the formation of a double stranded break in the gp17 gene. The break will be repaired with randomized oligos flanked by homologies to the cut site, leading to the formation of a T7 phage library.</figcaption>
</figure>
The main advantage of the yeast approach is the high flexibility in the choice of the target region. Once the system is established, it is relatively easy to change the site of mutagenesis: the modification comes down to the design of a novel gRNA and randomized oligos containing homologies to the new cut site. Additionally, it should be possible to target multiple regions at once. The library size can easily be increased by scaling up the yeast culture.
Furthermore, the yeast approach can be used to edit any species of bacteriophage. This is an advantage over our other two methods: the recombineering approach is dependent on whether the bacteriophage to be engineered can infect the recombineering strain that contains the donor plasmid. It is not guaranteed that for any bacteriophage a compatible strain is available. The in vitro approach is dependent on the presence of restriction sites at the site of interest. If these are missing, work is needed to introduce restriction sites at the locus of interest.
Stable and Selectable YAC-T7 Vector
To modify the T7 genome in yeast, it first has to be integrated in a stable and selectable manner in a yeast strain. This is achieved my means of our YAC-T7 vector, a plasmid that contains the T7 genome fused to an origin of replication and a selection marker for uptake and correct assembly. In order to ensure that only the cells that repaired the double stranded break in the T7 genome with the provided randomized oligos survive, a centromeric region (CEN) that allows for the one-copy propagation of the plasmid, was chosen as origin of replication. Even though the CRISPR/Cas9 system is extremely powerful, it does not guarantee that all target regions are cut in the case of a multi-copy plasmid. The presence of multiple copies of the T7 genome in a single yeast cell would allow for the homologous repair with copies of the non-cut wild type T7 genome and therefore lead to a higher false positive rate.
To select for repaired plasmids an auxotrophic marler is integrated into our YAC-T7 vector. S. cerevisiae BY4741 (genotype: MATa his3Δ1 leu2Δ0 met15Δ0 ura3Δ0) was selected as chassis because it allows for the selection of the auxotrophic marker his3p (HIS3). With the integration of HIS3 into our YAC-T7 vector we allow cells that perform the uptake and correct assebly to grow in minimal medium lacking histidine.
Primers that amplify the yeast artificial chromosome containing the CEN and HIS3 cassette were designed to contain overhangs to the ends of the linear T7 genome. The integration of NotI restriction sites allow us to cut the T7 genome back out of the YAC-T7 vector. The yeast backbone, taken from the pRG213MX shuttle vector [2]<a style="color: #ffffff; text-decoration:none;" href="#biblio-yeast">Ando, H., Lemire, S., Pires, D. P., & Lu, T. K. (2015). Engineering Modular Viral Scaffolds for Targeted Bacterial Population Editing. Cell systems, 1(3), 187–196. doi:10.1016/j.cels.2015.08.013</a>, is submitted as biobrick <a class="a-link" href="http://parts.igem.org/Part:BBa_K3211202">BBa_K3211202.</a> The YAC-T7 vector was assembled by gap repair of two fragments in yeast: the backbone with overhangs to the T7 genome and linear T7 genome itself. The correct assembly of the YAC-T7 vector was verified by PCR amplification of the junctions between the backbone and the T7 genome, and selected regions of the centromeric origin of replication (CEN), the selection marker (HIS3) and of the T7 genome (Figure 2). Sequencing confirmed the correct assembly and the presence of the NotI restriction sites within the junctions.
<figure class="figure-center">
<img src="
" alt="T7 assmbly check">
<figcaption> Figure 2: Assembly of a Stable and Selectable YAC-T7 Vector in Yeast. Primers containing overhangs to the T7 genome are used to PCR amplify the YAC backbone with the centromeric region and a selectable his3p gene from the pRG213MX shuttle vector [2]<a style="color: #ffffff; text-decoration:none;" href="#biblio-yeast">Gnügge, R. & Rudolf, F. Saccharomyces cerevisiae Shuttle vectors. Yeast 34, 205–221 (2017).</a>. The overhangs contain NotI restriction sites that allow us to cut the T7 genome out of the YAC-T7 vector to recover the modified genome and form phages in the next steps. The PCR amplified YAC and linear T7 genome were transformed into yeast and were assembled into the YAC-T7 vector by gap repair. PCR amplification of junction 1, junction 2, HIS3, CEN and T7 resulted in bands of the correct size and sequencing of the PCR products further verified the correct assembly of the YAC-T7 vector including the presence of the introduced NotI restriction sites.
</figcaption>
</figure>
Formation of Functional T7 Phages from the YAC-T7 vector
A crucial step in the engineering of bacteriophages in yeast is the ability to extract the altered genome and form functional phages. The size of 42 kb of the YAC-T7 vector makes its extraction challenging because most available commercial kits cannot be used for the efficient purification of large DNA. Therefore, several kit free methods were combined in order to purify the YAC-T7 vector. Two aspects, namely the high abundance of RNA in comparison to the one-copy YAC-T7 vector in yeast and the chemical similarity between RNA and DNA, led to a high ratio of RNA to DNA after the extraction from yeast. An RNase A digest was used to degrade the surplus of RNA. By choosing conditions that selectively precipitate large DNA <a class="a-link" href="https://static.igem.org/mediawiki/2019/9/9f/T--ETH_Zurich--cleanup-yeast-protocol.pdf">(Large Plasmid Extraction Cleanup),</a> the YAC-T7 vector was isolated from the remaining small RNA fragments (Figure 3A). Electroporation of the extracted YAC-T7 vector into E. coli led to the formation of plaques. This confirms that the introduced YAC-T7 vector can be extracted and used for the formation of functional phages. No plaques were formed when the same amount of extracted YAC-T7 vector was added to cells without electroporation which confirmed that the extracted DNA was not contaminated with phages (Figure 3B).
However, no functional phages could be formed in the cell-free transcription-translation system with the YAC-T7 vector. A possible reason could be that the carry-over of salts from the isopropanol precipitation inhibits the system. This hypothesis is supported by the fact that it was also not possible to digest the T7 genome with a restriction enzyme from the YAC, even though sequencing results clearly indicate the presence of the restriction sites. In a next step, it would be important to optimize the extraction protocol in order to make it compatible with cell free transcription-translation systems.
<figure class="figure-center">
<img src="
" alt="Plasmid extraction">
<figcaption>Figure 3: Rebooting of the Phage Genome by Electroporation into Bacteria. A) An RNase A digest followed by Potassium Acetate/Isopropanol precipitation efficiently purifies the YAC-T7 vector from RNA. Note: the RNA cloud present at the bottom of the RNase digest gel is not visible in the gel of the extraction because this gel ran for a longer time as can be seen from the ladder. B) The extracted YAC-T7 vector was electroporated into E. coli SW102 and plaque assays were made shortly after electroporation. The ability of the YAC-T7 vector to form functional phages was confirmed by the formation of plaques, corresponding to lyzed bacteria, when electroporated. No plaques were observed when bacteria were mixed with YAC-T7 vector without electroporation (negative control). This confirms that the YAC-T7 vector DNA was not contaminated with phages.</figcaption>
</figure>
Induction of a Double Stranded Break in the YAC-T7 Vector by CRISPR/Cas9
In order to introduce the library at the site of interest in the gp17 gene, a CRISPR/Cas9 system with a gRNA designed to target the BC loop in the gp17 gene was introduced into the strain. The expression of CRISPR/Cas9 is induced with β-estradiol [3]<a style="color: #ffffff; text-decoration:none;" href="#biblio-yeast"></a>Gnügge, R. & Rudolf, F. Saccharomyces cerevisiae Shuttle vectors. Yeast 34, 205–221 (2017).. The functionality of the CRISPR/Cas9 system was verified by testing the viability of cells with or without CRISPR/Cas9 induction on different selection media. If the system introduces a cut at the target site, the YAC-T7 vector is linearized leading to its degradation and the loss of the HIS3 selection marker. As a result, viability is decreased on medium lacking histidine.
Cells were grown to exponential phase and half of the cells were induced with β-estradiol for 1.5 hours in liquid culture. Dilution series (factor 10 for each dilution) were spotted onto complete synthetic defined media or synthetic defined medium lacking histidine (-HIS). Figure 4 illustrates that cells where the CRISPR/Cas9 system was induced only cut within the YAC-T7 vector and not in off targets within the yeast genome. Indeed, cells do not lose viability on complete synthetic defined media when induced (Figure 4, compare left bar and central bar). This confirms that the gRNA has no off-targets in the genome. When plated on plates with medium lacking histidine, viability is significantly decreased for the induced cells (Figure 4, central bar to right bar), which confirms that Cas9 cuts the YAC-T7 vector leading to its linearisation and to the loss of the HIS3 selection marker.
<figure class="figure-center">
<img src="
" alt="CRISPR/Cas9">
<figcaption>Figure 4: Correct functioning of the CRISPR/Cas9 and gRNA System. CRISPR/Cas9 was induced with β-estradiol for 1.5 hours in liquid culture. Dilution series (factor 10) of induced and non-induced cells were spotted on complete synthetic defined media and synthetic defined medium lacking histidine (-HIS). If Cas9 cuts at the designed position, the cells will lose the YAC-T7 vector and the histidine selection marker. Therefore, the viablilty should decrease on -HIS plates when Cas9 is induced. This exact effect was observed (central and right bars). The gRNA shows no off-target effects in the genome as can be deduced by the sustained cell viability on full growth medium (left and central bars).
</figcaption>
</figure>
Swapping T7 Tail Fiber Region by its T3 Homologue
After successfully establishing and testing all components needed for phage genome editing in yeast, we performed our gold standard experiment to test if the system can exchange the targeted region of the T7 gp17 tail fiber protein with its T3 homolog. To this end, the CRISPR/Cas9 system was induced for 1.5 h with β-estradiol prior to the transformation of a short oligo containing the T3 homolog flanked by T7 homologies to the cut site. In addition to the T3 oligo, a helper plasmid with an orthogonal selection marker was transformed. This allows for the selection of cells that have taken up DNA and decreases the false positive rate associated with cells that are able to repair the double stranded break by means other than homologous recombination with the provided oligo. Sequencing results confirmed the successful replacement of the targeted region with the provided T3 homolog (Figure 5). This shows that the yeast sytem can effectively be used in order to edit phage genomes.
<figure class="figure-center">
<img src="
" alt="sequenceing_t3">
<figcaption>Figure 5: Swap of a region of the T7 tail fiber protein with its T3 homolog. The perfect alignment of T3 wild type phage to the T7/T3 hybrid phage produced by homologous recombination in yeast confirms that the provided linear DNA template containing the T3 tail fiber homologue was effectively introduced at the designed position in the phage genome. By comparing the sequence of the modified genome in yeast ("T3/T7 Swap" on figure) to T7 wild type tail fiber protein we can observe the expected deletions.
</figcaption>
</figure>
Making Phage Libraries with the Yeast Approach
We showed that the yeast system can effectively be used to engineer a phage genome at the position of interest. Finally, we wanted to show, that it is possible to introduce a library at the site of interest by providing a randomized oligo with homology to the cut site. The randomized oligo is assembled by polymerase chain assembly (PCA) and detailed information on its design can be found <a class="a-link" href="#randomized_fragment">here</a>. It is the same fully randomized fragment that we used in the in vitro approach. Briefly, it contains four randomized regions, each being a couple of codons long, that are not conserved between homologs of the tail fiber proteins and are therefore believed to influence host specificity. The regions are called BC, DE, FG and HI loop.
Figure 6 shows the sequencing results of two colonies for which the library has been successfully introduced into the BC and the HI loop, respectively. For the first colony, solely the BC loop was randomized. Considering the fact that the degenerate oligo is only randomized in selected regions, homologous recombination is possible in between the randomized regions leading to only the randomization of the BC loop. This problem could be avoided by introducing a second gRNA that targets the terminal HI loop and would lead to the complete loss of the fragment that is to be replaced by the degenerate oligos. Is has recently been shown that it is in fact possible to successfully introduce up to 25 gRNAs into a CRISPR system [4]<a style="color: #ffffff; text-decoration:none;" href="#biblio-yeast">Campa, C. C., Weisbach, N. R., Santinha, A. J., Incarnato, D., & Platt, R. J. (2019). Multiplexed genome engineering by Cas12a and CRISPR arrays encoded on single transcripts. Nature methods, 16(9), 887-893.</a>. The sequencing of the second colony shows a new sequence in the HI loop. Unfortunately, all other loops only show unassigned Ns. This could be due to the fact that colony PCR was performed by directly picking the selected colonies from the plate and therefore contamination with degenerate oligos added during the transformation but which were not introduced into the T7-YAC vector could have taken place. In total, 10 colonies were sequenced, but only the two clear sequences shown in Figure 5 could be obtained due to the mentioned contamination. Sequencing will be repeated by picking single colonies from a freshly streaked plate.
The transformation efficiency was dependent on the used helper plasmid. Three different helper plasmids were tested and lead to 0 colonies, roughly 50 colonies or to a fully grown plate. The choice of the helper plasmid as well as the exact transformation conditions will need to be optimized and the exact transformation efficiency needs to be properly estimated.
<figure class="figure-center">
<img src="
" alt="Library-sequencing">
<figcaption>Figure 6: Sequencing Results of the Randomized YAC-T7 Vector. The sequencing results confirm the introduction of novel sequences at the designed positions.
</figcaption>
</figure>
To conclude, it was possible to establish all essential elements needed for the engineering of phages in yeast. A stable and selectable YAC-T7 vector was successfully introduced into the system. A gRNA was designed that upon induction of the CRISPR/Cas9 system leads to the cut at the target position with no observable off-target effects. It was possible to randomize the positions of interest. Finally, successful rebooting of the phage genome by electroporating the YAC-T7 vector into bacteria shows that the yeast system can effectively be used in order to produce phage libraries.
Bibliography
[1] Ando, H., Lemire, S., Pires, D. P., & Lu, T. K. (2015). Engineering Modular Viral Scaffolds for Targeted Bacterial Population Editing. Cell systems, 1(3), 187–196. doi:10.1016/j.cels.2015.08.013
[2] Gnügge, R. & Rudolf, F. Saccharomyces cerevisiae Shuttle vectors. Yeast 34, 205–221 (2017).
[3] Ottoz, D. S. M., Rudolf, F. & Stelling, J. Inducible, tightly regulated and growth condition-independent transcription factor in Saccharomyces cerevisiae. Nucleic Acids Res. 42, e130 (2014).
[4] Campa, C. C., Weisbach, N. R., Santinha, A. J., Incarnato, D., & Platt, R. J. (2019). Multiplexed genome engineering by Cas12a and CRISPR arrays encoded on single transcripts. Nature methods, 16(9), 887-893.
</div>
Recombineering
Design Overview
<p>Our recombineering approach uses the homologous recombination machinery of E. coli cells to insert a DNA sequence from a plasmid into the phage’s genome. The plasmid contains random codons flanked by sequences homologous to the phage genome. During the infection by a phage, homologous recombination can occur between the plasmid and the phage, resulting in the replacement of a piece of the phage genome by the sequence in the plasmid.</p> <p>
</p> <figure class="figure-center"> <a href=""><img src="
</p> <p>This can easily be scaled up due to the low costs of the necessary reagents. However, the homologous recombination is not very efficient and most of the phages produced will keep their original tail fibers, thus reducing the library size. Nevertheless, this can be fixed by scaling up sufficiently.</p>
Phage genome editing by homologous recombination
<p>In order to find out if T7 genome editing is possible via in vivo homologous recombination with a donor plasmid, we attempted to replace different parts of the T7 tail fiber with the T3 homologue. T7 and T3 have very similar genomes and their tail fibers are highly conserved. Unconserved sequences in the tail fibers were selected (Fig. 2). E. coli harboring corresponding donor plasmids were infected with T7. In the resulting lysates, the presence of recombined phages was checked by PCR. For this a sequence was amplified by primers that bind to the genome inside and outside of the swapped sequence (Fig. 3).
</p> <p>
</p> <figure class="figure-center"> <img src=""> <figcaption>Figure 2: Selected unconserved regions between the T7 and T3 phage. At these loci the T3 sequence was inserted into the T7 phage by homologous recombination. In the rightmost case all of the individual regions were swapped at once.</figcaption> </figure> <p>
</p> <figure class="figure-center"> <img src="" alt="gel recombineering"> <figcaption>Figure 3: PCR demonstrating successful recombination. We attemped to replace different parts of the T7 phage's tail fiber gene with the T3 homologue. Successful replacement was verified by PCRs with one primer that binds to an unswapped region of the T7 genome and another that binds inside the swapped sequence. In the negative controls primers were added that should amplify if one of the other swaps occurred. All four swaps were also done at once as one big swap (we have no negative control here as all the swaps happen). As a positive control, the PCRs were performed with the T3 genome as a template, which should give approximately the same band lengths, because the primer binding outside of the swapped regions binds to a sequence conserved between T7 and T3.</figcaption> </figure> <p>
</p>
Host range alteration by tail fiber editing
<p>It has been shown that by changing the T7 tail fiber to the T3 homologue, the host range can be altered. The E. coli ECOR16 strain is not infected by T7 but is by T3. However, by changing the C-terminal domain of the tail fiber of the T7 phage to the T3 homologue, the resulting hybrid phage can infect [1]. We successfully reproduced this swap in infectivity (Fig. 4), and attempted to find which amino acids were responsible for the specificity. For this, all unconserved parts were swapped individually (Fig. 2) and the infectivities toward ECOR16 were measured. We found that only the hybrid with the entire C-terminal domain of T3 is able to infect ECOR16 (Fig. 5). A plaque was picked and the tail fiber region was sequenced. It had precisely the expected recombined sequence.</p> <p>
</p> <figure class="figure-center-50"> <img src=""> <figcaption>Figure 4: Plaque assay of T7 phages with the T3 C-terminal tail fiber domain on E. coli ECOR16.</figcaption> </figure> <p>
</p> <figure class="figure-center-50"> <img src="https://static.igem.org/mediawiki/2019/d/d1/T--ETH_Zurich--swaps-infectivities.svg"> <figcaption>Figure 5: Infectivity of different T7-T3 hybrids. Different unconserved regions of the T7 tail fiber were swapped for the T3 variant (Fig. 2). These were used to do plaque assays on E. coli ECOR16, which is infected by T3 but not by wild type T7.</figcaption> </figure> <p>
</p>
Plasmid library generation
<p>A library of donor plasmids is generated by PCR amplifying a plasmid that consists of a backbone and a fragment of the phage genome. Primers with degenerate bases are used, such that the linear PCR product has homologous ends, and at one end a short stretch of bases in the phage genome fragment are replaced by random codons. This PCR product is circularized by Gibson assembly and transformed into E. coli (Fig. 6). This results in a library of bacterial colonies, each carrying a different donor plasmid.</p> <p>
</p> <figure class="figure-center"> <img src="" alt="plasmid-library-generation"> <figcaption>Figure 6: Generation of the plasmid library. A plasmid with a fragment of the phage genome containing the sequence to be randomized is amplified with a normal primer and a primer containing degenerate bases such that a linear product with homologous end is created, which can recircularize to yield a donor plasmid. These donor plasmids are transformed into E. coli and give a library of colonies, each carrying a different donor plasmid.</figcaption> </figure> <p>
</p> <p>The transformation with the library of donor plasmids yielded 10^4 colonies. To ensure that indeed all colonies carried different donor plasmids, 10 were picked and sequenced. 9 carried a unique sequence with degenerate bases at the expected positions, while 1 had an unexpected insertion. Therefore 90% of the colonies carried unique and functional donor plasmids (Fig. 7).</p> <p>
</p> <figure class="figure-center"> <img src=""> <figcaption>Figure 7: Sequencing of our colony library. From the library of colonies carrying different donor plasmids, 10 were picked and sequenced to ensure that the donor plasmids were indeed unique. 1 of the 10 sequences had an unexpected insertion, while the other 9 all carried unique donor sequences with degenerate bases at the expected positions. The sequencing data of these 9 donor plasmids is shown at the region where the random bases are inserted.</figcaption> </figure> <p>
</p>
Phage library generation
<p>To generate a library of phages, the library of colonies is pooled, grown in liquid culture and then infected by phages. Given a fraction of 10-3 recombinant phages, a final phage titer of 1010mL-1 in the lysate and a colony library size of 104, it can be assumed that the entire colony library is covered by the phage library.</p>
Construction of E. coli resistant to T7 with a CRISPR-Cas system targeting the tail fiber
<p>In order to validate the presence of the phage library, given our low recombination efficiency of 10^-3, it would be necessary to screen at least thousands of plaques for library members, but probably even more, because on average library members are expected to be less infective than wild type phages. In order to be able to select for library members, E. coli with CRISPR-Cas targeting the non-randomized wild type phage genome was constructed. It was confirmed by spot assays that cells with the CRISPR-Cas system was infected 10^5 fold less efficiently by T7 than cells without it (Fig. 8).</p> <p>
</p> <figure class="figure-center"> <img src=""> <figcaption>Figure 8: Effectiveness of the CRISPR-Cas system against T7. Infectivity of T7 on wild type E. coli DH5alpha cells and cells engineered with a CRISPR-Cas system targeting T7.</figcaption> </figure> <p>
</p>
Library contains variants capable of infecting CRISPR-resistant bacteria
<p>To isolate single phages from the library, plaque assays were performed on the cells with CRISPR-Cas targeting wild type T7. 7 single plaques were picked and the region of the genome containing the library insert was sequenced. 5 different sequences were found (Fig. 9), demonstrating that we have obtained a library of phages, including variants that retain their infectivity towards E. coli DH5alpha.</p> <p>
<p> <figure class="figure-center"> <img src=""> <figcaption>Figure 9: Sequences of plaques originating from our phage library. Library members were selected for by plating on cells with a CRISPR-Cas system targeting wild type T7 but not library members. On the bottom the sequence of T7 is shown as a reference.</figcaption> </figure>
Selection of good variants with the bioreactor
In vitro
Design Overview
<p>
The in vitro approach focuses on generating the T7 library using cell-free methods. It is based on the creation of three fragments, which are ligated to form the T7 genome with a randomized gp17 sequence. The new T7 DNA is used to produce bacteriophages in a cell-free system (TXTL) circumventing the transformation step, which would be inefficient due to the large size of the T7 genome. To ensure the absence of wild-type T7 phages in our library, the T7 DNA is digested with a restriction enzyme what leads to a blunt end cut. This cannot be re-ligated by the T7 DNA ligase and functional phages form only when this fragment is replaced with an uncut DNA piece (fragment 2+3). Compared to the other approaches, this method has the advantage of producing 100% phages with novel tail fibers but is limited in terms of scale-up.
</p>
<p>
</p>
<figure class="figure-center">
<img src="
" alt="Overview of the in vitro approach">
<figcaption>Figure 1: Overview of the in vitro approach. Three fragments are generated using restriction digests, polymerase cycling assembly and polymerase chain reactions. Fragments are ligated with T7 DNA ligase. A cell-free system is used to generate phages with a randomized tail fiber protein. </figcaption>
</figure>
<p>
</p>
Generation of fragment 1
<p>
Fragment 1 is generated by restriction digests, using the T7 genome as a template. The mechanism can be seen in figure 2 and the completely digested T7 DNA is shown in figure 3.
</p>
<p>
</p>
<figure class="figure-center-50">
<img src="
" alt="Method to generate fragment 1">
<figcaption>Figure 2: Method to generate fragment 1. The T7 genome is digested with the restriction enzymes SfiI and PmlI. SfiI creates an overhang that is used for sticky-end ligation to the randomized fragment 3. The resulting fragment is 36 kb in length.
</figcaption>
</figure>
<p>
</p>
<figure class="figure-center-50">
<img src="
" alt="Verification of restriction digests">
<figcaption>Figure 3: Verification of restriction digests. The complete digest of T7 DNA with the restriction enzymes SfiI and PmlI could be verified using gel electrophoresis..</figcaption>
</figure>
<p>
</p>
Generation of fragment 2
<p>
For fragment 2 a PCR approach is used. The forward primer introduces a type IIS restriction enzyme binding site for the enzyme Esp3I and with the reverse primer the downstream region of the T7 DNA is covered. The PCR product digested with Esp3I forms a sticky-end compatible with the BtgI digested fragment 3, Fig. 4. The length of the generated fragment was confirmed in an agarose gel, Fig. 5.
</p>
<p>
</p>
<figure class="figure-center">
<img src="
" alt="Method to generate fragment 2">
<figcaption>Figure 4: Method to generate fragment 2. (a) The T7 genome was digested with SfiI. The 4 kb fragment was used as a template for a PCR reaction introducing a type IIS restriction enzyme binding site through the forward primer. (b) Digestion by Esp3I leads to an overhang compatible with the overhang of the randomized fragment 3.
</figcaption>
</figure>
<p>
</p>
<figure class="figure-center-50">
<img src="
" alt="Validation of the length of fragment 2">
<figcaption>Figure 5: Validation of the length of fragment 2. The expected length of 3622 bp was validated by gel electrophoresis.
</figcaption>
</figure>
<p>
</p>
Generation of the randomized fragment 3
<p>
Fragment 3 encodes mostly the surface structures of the tail fiber protein. We randomized the sequence at four loop structures, using an adapted version of the PCR-based two-step DNA synthesis (PTDS) method described by Xiong et al. [1]<a style="color: #ffffff; text-decoration:none;" href="#biblio-in-vitro">Xiong AS, Yao QH, Peng RH, Li X, Fan HQ, Cheng ZM, Li Y. A simple, rapid, high-fidelity and cost-effective PCR-based two-step DNA synthesis method for long gene sequences. Nucleic Acids Res, 32(12), 2004</a>. We chose these loops as randomization targets, as they stick out of the protein surface and are important for binding to bacterial receptors and host specificity [2]<a style="color: #ffffff; text-decoration:none;" href="#biblio-in-vitro">Garcia-Doval C, van RaaiJ MJ. Structure of the receptor-binding carboxy-terminal domain of bacteriophage T7 tail fibers. PNAS, 109(24), pp. 9390-9395, 2012</a>.
</p>
<figure>
<p>
</p>
<figure class="figure-center">
<img src="
" alt="Position loop structure T7">
<figcaption>Table 1: Position of loop structure in T7 genome. Position is indicated in base pairs. </figcaption>
</figure>
<p>
</p>
<p>
The oligonucleotides used in the PTDS method are on average 60 bp long and are designed to have 20 bp overlaps. For the loop structures, both degenerate oligos as well as the original sequences were ordered, allowing for both separate and combinatorial randomization.
In a first reaction, the oligos anneal to complementary fragments and DNA polymerase fills up the gaps. The second reaction uses outside primers to only amplify complete fragments. As this step will only lead to the amplification of existing fragments and not the enlargement of our library, the number of PCR cycles were reduced compared to the PTDS protocol from Xiong et al. An overview of the methods is shown in Fig. 6. The integration of the randomized oligonucleotides was verified by sequencing as shown in Fig. 7. Introduced N’s in the defined sequences can be explained by the possibility of deletions in the ordered oligonucleotide sequences. To enable ligation to the fragments 2 and 3, compatible overhangs were created by restriction digests with SfiI and BtgI, Fig. 8.
</p>
<p>
</p>
<figure class="figure-center">
<img src="
" alt="Method to generate the randomized fragment">
<figcaption>Figure 6: Method to generate the randomized fragment 3. The randomized fragment is constructed by the PTDS method using 60 bp long oligonucleotides with a 20 bp overlap. Degenerate oligos are used for the loop structures. The generated fragments are further amplified by PCR. Compatible overhangs with fragment 1 and 2 are created by restriction digests with SfiI and BtgI.
</figcaption>
</figure>
<p>
</p>
<p>
</p>
<figure class="figure-center">
<img src="https://static.igem.org/mediawiki/2019/4/41/T--ETH_Zurich--invitro-verif-rand-sanger.svg" alt="Verification of randomization by Sanger sequencing">
<p>
</p>
<figcaption>Figure 7: Verification of randomization by Sanger sequencing. Sequencing shows the successful randomization of the loop structures.</figcaption>
</figure>
<p>
</p>
<figure class="figure-center-50">
<img src="
" alt="Verification of restriction digests">
<figcaption>Figure 8: Verification of restriction digests. The expected length of the undigested fragment 3 (F3) is 427 bp, the single-digest with SfiI creates a 402 bp fragment and the double-digest with Sfi and BtgI results in a 377 bp fragment.
</figcaption>
</figure>
<p>
</p>
<p>
The beauty of this method is, that the loops can be randomized both simultaneously and all combinations without additional designing effort. This can be utilized in future experiments to explore the importance of each loop for host specificity.
A disadvantage remains in the fact, that large oligonucleotides can currently only be synthesized with a success rate of close to 70% for 60 bp, meaning that the remaining 30% have deletions. When assembling multiple oligonucleotides this error increases. This leads to T7 genomes that contain a frameshift in the gp17 gene, causing non-functional tail fiber proteins. The decrease in resulting plaque forming units was estimated by ligating once a preordered gBlock (IDT) and once the assembled non-randomized product generated with the PTDS method (see Fig. 9). The result shows that the PTDS method reduces the library size by up to a 100-fold.
</p>
<figure class="figure-center-50">
<img src="
" alt="comparing pfu ligation gBlock PTDS">
<figcaption>Figure 9: Comparing the PFU from ligation with a gBlock and the product of the PTDS method. The T7 DNA was ligated using once a gBlock as template for fragment 3 and once the PTDS method. The DNA was used to form phages in TXTL and the plaque forming units (PFU) were calculated on DH5alpha.
</figcaption>
</figure>
<p>
</p>
Ligation
<p>
The three fragments are ligated using T7 DNA ligase as shown in Fig. 10. First, fragment 2 and 3 are ligated. We optimized this step by testing different molar ratios of the fragments (Fig. 11 (a)). We chose a molar ratio of 1:5. In a second ligation reaction, the pre-ligated fragment 2+3 is combined with fragment 1. A molar ration of 1:4 was used to outcompete the wild type fragment, Fig. 11 (b).
</p>
<p>
</p>
<figure class="figure-center">
<img src="
" alt="Ligation strategy">
<figcaption>Figure 10: Ligation strategy. First the randomized fragment 3 is ligated with fragment 2 in a 5:1 molar ratio. The gel-purified ligation product is then ligated with fragment 1 in a 1:4 molar ratio. Functional phages are produced only when the pre-ligated fragment 2+3 is ligated.
</figcaption>
</figure>
<p>
</p>
<figure class="figure-center">
<img src="
" alt="Testing of ligation efficiencies">
<figcaption>Figure 11: Testing of ligation efficiencies. (a) Different ratios of the randomized fragment and fragment 2 were tested for an optimal ligation efficiency. The efficiencies were measured using gel electrophoresis. The lower band shows non-ligated fragment 2, whereas the upper band indicates the successful ligation of fragment 2 and 3. (b) Different molar ratios of fragment 1 and the pre-ligated fragment 2+3 were ligated to optimize the phage formation. The efficiencies were determined by adding ligated T7 DNA into the cell-free expression mix (myTXTL Arbor Biosciences). By performing plaque assays, the number of plaque forming units (PFU) were determined using DH5alpha as host bacterium.
</figcaption>
</figure>
<p>
</p>
Phage formation
<p>
</p> <p> In a cell free system, gene transcription (TX) and translation (TL) is executed in a single reaction tube utilizing the TXTL machinery of bacteria. Gene expression is initialized by adding T7 template DNA and phages are formed as shown by Shin et al. [3]<a style="color: #ffffff; text-decoration:none;" href="#biblio-in-vitro">Shin J, Jardine P, Noireaux V. Genome Replication, Synthesis, and Assembly of the Bacteriophage T7 in a Single Cell-Free Reaction. ACS Synth. Biol., 1(9), pp. 408-413, 2012</a> as well as last year's iGEM team from Munich. The concept is illustrated in figure 12. In this project the cell free expression system for linear DNA from Arbor Biosciences (myTYTL – linear DNA expression kit, TXTL) was used. </p> <p>
</p> <figure class="figure-center"> <img src="" alt="concept phage formation cell-free"> <figcaption>Figure 12: The concept of phage formation in a cell-free expression systemDNA is replicated and transcribed into mRNA, which is further translated into proteins leading to the self-assembly of a functional phage with packaged DNA. </figcaption> </figure> <p>
</p> <p> As described in the overview section the formation of wild-type phages can be circumvented by introduction of an additional blunt end cut in the phage genome that cannot be ligated by the T7 ligase. The restriction enzyme PmlI was used as it cuts in the essential gene 19, which is needed for DNA maturation [4]<a style="color: #ffffff; text-decoration:none;" href="#biblio-in-vitro">Dunn JJ, Studier W, Gottesman M. Complete nucleotide sequence of bacteriophage T7 DNA and the locations of T7 genetic elements. JMB, 16(4), pp. 477-535, 1983</a>. This blunt end cannot be re-ligated with the T7 DNA ligase and leads to non-functional phages. Only when the fragment is replaced with the pre-ligated fragment 2+3, functional phages are formed, Fig. 13. This strategy was experimentally confirmed as it can be seen in figure 14. This indicates that 100% of the formed phages are novel using this method. It also shows that the fragment 3 generated with the PTDS method forms functional phages and that a phage library can be created using the same method. </p> <p>
</p> <figure class="figure-center"> <img src="" alt="strategy avoid wild-type"> <figcaption>Figure 13: Strategy to avoid wild-type phage formation. A schematic representation of how the blunt end cut in the T7 DNA leads non-functional phages. </figcaption> </figure> <p>
</p> <figure class="figure-center-50"> <img src="" alt="testing phage formation txtl"> <figcaption>Figure 14: Testing phage formation in TXTL. Wild-type T7 DNA was digested once with PmlI only and twice with PmlI and SfiI. After the re-ligation with the T7 DNA ligase and expression in TXTL, no functional phages were formed when the fragment 2+3 was not added into the ligation reaction (w/o F2+3). With addition of the wild-type fragment 2+3 into the ligation reaction (w F2+3), functional phages were formed. </figcaption> </figure> <p>
</p>
The random adhesion of tail fiber proteins
<p>
</p> <p> For the expression of our DNA library in TXTL, an additional challenge has to be overcome. The transcribed mRNAs will most likely diffuse away from their DNA and therefore, the genome will not be packaged with the right tail fiber proteins. This results in a loss of the genotype-phenotype linkage. If a phage has the specific tail fiber proteins to infect a novel host, it can infect only once. The new tail fiber produced in the second round will not be specific for the host as illustrated in Fig. 15. </p> <p>
</p> <figure class="figure-center"> <img src="" alt="testing phage formation txtl"> <figcaption>Figure 15: The tail fiber protein problem. In TXTL mRNA can diffuse away from the DNA leading to phage with non-matching tail fiber proteins and genomes. Overexpressing tail fiber proteins that can infect a known bacterium is needed as an intermediate step to achieve protein and genomic compatibility. </figcaption> </figure> <p>
</p> <p> Three approaches were tested to overcome this problem:
First, purified wild-type tail fiber proteins were added directly into the TXTL mix in order to outcompete randomized tail fibers. However, phage formation was reduced 100-fold as the protein buffer is inhibiting the TXTL reaction as seen in Fig. 16. This was observed even though the <a class="a-link" href="https://static.igem.org/mediawiki/2019/c/cd/T--ETH_Zurich--strep_tag_collection_manual.pdf">protein buffer</a>was prepared with ions that should not inhibit the TXTL machinery (according to the manufacturers specifications). </p> <p>
</p> <figure class="figure-center-50"> <img src="" alt="influence protein buffer"> <figcaption>Figure 16: Influence of protein buffer on phage formation in TXTL. The commercial T7 DNA was resuspended once in water and once in protein buffer. Both solutions were added into the TXTL solution and the phage formation (PFU) was determined using DH5alpha as bacterial host. </figcaption> </figure> <p>
</p> <p> Second, DNA encoding for the wild-type tail fiber protein can be added to TXTL. Therefore, a DNA sequence was designed containing an Anderson promoter, the tail fiber gene sequence and a terminator. We are currently cloning the sequence into a suitable vector for protein expression in the TXTL.
A third option is to directly electroporate the DNA into bacteria. However, the electroporation is rather inefficient for large DNA fragments and even more for linear DNA [5]<a style="color: #ffffff; text-decoration:none;" href="#biblio-in-vitro">Sheng, Y., Mancino, V., & Birren, B. (1995). Transformation of Escherichia coli with large DNA molecules by electroporation. Nucleic acids research, 23(11), 1990-1996.</a>. No infecting phages could be isolated when the phage DNA library was electroporated into a bacterial host. However, this does not exclude the possibility of a present phage library and its existence has to be verified in additional experiments. </p> <p>
</p>
Swapping tail fibers
<p>
</p> <p> To check whether the host specificity can be swapped with this method, the tail fiber surface of the T7 phage was exchanges against that of the T3 phage. For this the fragment 3 which encodes mostly for the surface structure of the T7 tail fiber protein was exchanged to the T3 sequence. Infecting the E. coli strain EcoR16 which can only be infected by the T3 phage showed that the host specificity had been altered, as can be seen in Fig. 17. Sequencing confirmed that the tail fiber region had been exchanged successfully. </p> <p>
</p> <figure class="figure-center-50"> <img src="
Bibliography
<p class="bibliography" id="biblio-in-vitro">
[1] Xiong AS, Yao QH, Peng RH, Li X, Fan HQ, Cheng ZM, Li Y. A simple, rapid, high-fidelity and cost-effective PCR-based two-step DNA synthesis method for long gene sequences. Nucleic Acids Res, 32(12), 2004
[2] Garcia-Doval C, van RaaiJ MJ. Structure of the receptor-binding carboxy-terminal domain of bacteriophage T7 tail fibers. PNAS, 109(24), pp. 9390-9395, 2012
[3] Shin J, Jardine P, Noireaux V. Genome Replication, Synthesis, and Assembly of the Bacteriophage T7 in a Single Cell-Free Reaction. ACS Synth. Biol., 1(9), pp. 408-413, 2012
[4] Dunn JJ, Studier W, Gottesman M. Complete nucleotide sequence of bacteriophage T7 DNA and the locations of T7 genetic elements. JMB, 16(4), pp. 477-535, 1983
[5] Sheng, Y., Mancino, V., & Birren, B. (1995). Transformation of Escherichia coli with large DNA molecules by electroporation. Nucleic acids research, 23(11), 1990-1996.
</p>
Phage training
<p>
An alternative to improve a phage’s infectivity for a specific host is phage training. Bacteriophages and bacteria coexist since billions of years on our planet what was only possible through co-evolution. The fluctuating selection scenario between the two species can be used to evolve a phage towards a desired host bacterium [1]<a style="color: #ffffff; text-decoration:none;" href="#biblio-phage-training">Rohde, C., Resch, G., Pirnay, J. P., Blasdel, B., Debarbieux, L., Gelman, D., ... & Łobocka, M. (2018). Expert opinion on three phage therapy related topics: Bacterial phage resistance, phage training and prophages in bacterial production strains. Viruses, 10(4), 178.</a>. In phage training, a bacterial liquid culture is infected by different dilutions of the phage. After 16 – 24 hours of incubation the sample with the lowest phage concentration in which the bacteria were lysed is selected. Phages from this sample are extracted, diluted and the host is infected again. This process is repeated several times until an optimal phage has evolved [2]<a style="color: #ffffff; text-decoration:none;" href="#biblio-phage-training">Merabishvili, M., Pirnay, J. P., & De Vos, D. (2018). Guidelines to compose an ideal bacteriophage cocktail. In Bacteriophage Therapy (pp. 99-110). Humana Press, New York, NY.</a>.
We infected a bacterial culture (E. coli DH5alpha) with different phages and measured the optical density of the bacterial culture over 40 hours using a plate-reader. The results are shown in figure 1 together with a negative control without phages (Fig. 1A). Infecting the culture with T7 phage lead to direct death of the host and no resistant bacteria were observed (Fig. 1B). Infecting the culture with T3 phage shows the phenomenon of resistant bacteria evolving with an increase in optical density (Fig. 1C). Usually, this is caused by a mutation in a gene which is required for the lipopolysaccharide’s assembly [3]<a style="color: #ffffff; text-decoration:none;" href="#biblio-phage-training">Perry, E. B., Barrick, J. E., & Bohannan, B. J. (2015). The molecular and genetic basis of repeatable coevolution between Escherichia coli and bacteriophage T3 in a laboratory microcosm. PLoS One, 10(6), e0130639.</a>. A phage might overcome this resistance through mutations in the tail fiber protein and regain its infectious potential [4]<a style="color: #ffffff; text-decoration:none;" href="#biblio-phage-training">Yehl, Kevin, et al. "Engineering phage host-range and suppressing bacterial resistance through phage tail fiber mutagenesis." Cell 179.2 (2019): 459-469.</a>. A similar observation was made with our self-generated T7 phage library (Fig. 1D). Firstly, the bacteria got lysed by the T7 phage which co-exists in the phage library. After several hours the optical density increased, indicating growth of resistant bacteria. Then the optical density decreased again, which is either caused by the T7 phage which co-evolved or a variant of our phage library that can efficiently infect the mutated bacteria. Further research needs to be done to verify one of the two possibilities.
</p>
<figure class="figure-center">
<img src="
">
<figcaption>Figure 1: Phage Training. A liquid DH5alpha culture was challenged with different phages and the optical density was measured over 4 hours at 37°C. The cells were grown without the influence of a phage (A), with T7 phage (B), T3 (C) and the T7/T3 hybrid (D). The multiplicity of infection (MOI) was 0.001 for the cultures infected with T7 and T3 and 0.01 for the phage library. All measurements are performed in three replicates (red, green and blue). </figcaption>
</figure>
Bibliography
<p class="bibliography" id="biblio-phage-training">
[1] Rohde, C., Resch, G., Pirnay, J. P., Blasdel, B., Debarbieux, L., Gelman, D., ... & Łobocka, M. (2018). Expert opinion on three phage therapy related topics: Bacterial phage resistance, phage training and prophages in bacterial production strains. Viruses, 10(4), 178.
[2] Merabishvili, M., Pirnay, J. P., & De Vos, D. (2018). Guidelines to compose an ideal bacteriophage cocktail. In Bacteriophage Therapy (pp. 99-110). Humana Press, New York, NY.
[3] Perry, E. B., Barrick, J. E., & Bohannan, B. J. (2015). The molecular and genetic basis of repeatable coevolution between Escherichia coli and bacteriophage T3 in a laboratory microcosm. PLoS One, 10(6), e0130639.
[4] Yehl, Kevin, et al. "Engineering phage host-range and suppressing bacterial resistance through phage tail fiber mutagenesis." Cell 179.2 (2019): 459-469.
</p>
</div>
</div>
</div>
</div>