As an alternative to the yeast-based genome assembly and TXTL approach, a novel approach using
recombination and CRISPR-Cas was used to engineer phage lambda. This method is based on recent work on
E. coli phage T4 and listeria phages [1, 2]. Parallel to this, we worked on hybrid phages,
which rely
on plasmid-based expression of the fusion protein. These can be used to test the chitin binding
properties of the engineered phage. The phage lambda engineering method consists of a recombination and a counterselection step. Methods in which CRISPR-Cas cutting is combined with rescue by recombination have been shown to be effective. [1, 2]
Our method separates the recombination and the selection in two different E. coli strains
and two different experiments. The first strain harbors two plasmids, one contains the lambda red system for recombination and the other one a recombination module to obtain recombined phages.
This strain is infected with phage lambda to obtain a recombinant phage lysate.
The second E. coli strain contains a Cas12a counterselection system to select against phages that have not recombined. Cas12a will recognize protospacers on the lambda genome and cut the phages that did not recombine.
Phages that did recombine will escape the counterselection and form pickable plaques. The recombination of phages is carried out in E. coli DH10B containing the lambda red system, which
increases recombination, and a plasmid containing the fusion protein flanked by 200 bp homology regions.
The fusion protein is gpD, which is a phage lambda capsid protein, fused to either the chitin binding
domain of Chitinase A1, chitin binding domain HAD, chitin binding protein PD1764 or GFP. These fusions
are all connected by the same flexible linker (CGSGSGSG). The flanking homology regions are
designed in such a way that the insert fusion protein is built in directly behind the original capsid
protein, as can be seen in figure 2. The choice was made for 200 bp homology regions as a good
compromise between ease of cloning and high recombination efficiency [3]. The recombinant phages are
generated by infecting the cells carrying the recombination system with phage lambda. This generates a lysate, from which the phages can be purified. The insertion constructs can be found as parts in the registry.
The counterselection is based on Cas12a. This endonuclease cuts double stranded DNA
(dsDNA) based on recognition of a 5’- TTN PAM and a 20 bp protospacer. As shown in figure 3, three
different spacers target the region that is replaced by successful recombination. Multiple spacers are
used to prevent low efficiency by an underperforming spacer. Counterselection DH10B E. coli
cells carry
a plasmid coding for Cas12a and a plasmid carrying one of these spacers. Counterselection is carried out by infecting the counterselection cells with the phage stock from the
recombination step. Recombinant phages escape the counterselection as the protospacer is removed by the
recombination. Phages that can escape the counterselection will form plaques that can be screened
through PCR. For the counterselection all three spacers are used individually. To decrease the chance of
phages escaping selection we also test a combination of the three spacers by doing a plaque assay with
mixed bacterial cultures. The mix of spacers should only show plaques for phages that can escape all
three spacers. This makes false positives less likely and reduces screening workload. To be able to quickly test the chitin binding of the phages, we worked on the hybrid phage system. By
expressing the capsid fusion on a plasmid and infecting cells with phages or by producing phages using a
Transcription Translation (TXTL) mixture, phages can be created that have a mix of WT and fusion
proteins on their capsid. The phages carrying fusion proteins can then be used in chitin binding assays
to determine the effectivity of the chitin binding. The hybrid phage system was constructed for both
phage lambda and phage T7, as production of T7 using TXTL has already been shown [4]. The hybrid phage
expression plasmid consists out of either lambda gpD or T7 capsid protein P10 fused by a flexible linker
(CGSGSGSG) to the different chitin binding proteins (Chitinase A1, HAD, PD1764) or GFP. The gpD fusions
are under the control of the Anderson J23116 promoter and the P10 fusions are under control of the T7
promoter. The J23116 is a weak Anderson promoter at 16% strength (relative to J23100). This choice was
made to avoid toxicity problems during cloning after earlier problems with J23100. For P10 the choice
was made for the T7 promoter as it was already in front of the gene on the T7 genome and could thus be
cloned in easily. Furthermore, the promoter is highly active when T7 infects the cell, giving expression
exactly when it is required. Using protein modelling we created a 3D model of the gpD to GFP fusion including a flexible linker (CGSGSGSG), which can be seen in figure 5.
In this model, it is observable that the gpD protein can correctly fold and is not sterically hindered by the fused domain.
The linker creates a clear separation between gpD and GFP.
This model gives us an indication that GFP could be correctly displayed on the phage capsid.
All the plasmids needed for the aforementioned experiment were successfully constructed, except for T7
P10+ChitinaseA, T7 P10+GFP and multiplexing counterselection spacer. These constructs suffered from
mutations and were abandoned due to lack of time. The recombination step was carried out according to the aforementioned design. To compensate for
potentially low level of recombinants, the counterselection step was carried out at a Multiplicity Of
Infection (MOI) of 1.5. The resulting plates did not show growth of E. coli. We hypothesized that the
phage concentration was so high that it allowed for many escape mutants or that it overwhelmed the
counterselection system. To correct for this, we repeated the counterselection with dilutions of 10^-3,
10^-4 and 10^-5 of the recombinant phage stock and performed spot assays with WT phage on the selection
cells. As can be seen in figure 4, the resulting plates still showed almost complete lysis of E. coli,
while the spot assays in figure 5 show no resistance of the cells. At this point we had to assume that
either our counterselection system was broken, which had been successfully tested on GFP and GFP before,
or that there was something wrong with our phage stock. Before the experiment started, the phage genome had been isolated using phenol chloroform method and had
been tested using digestion, with restriction enzymes EcoRI and PstI, and PCR to confirm the identity of
the phage. Whilst the digestion did not show bands except for a band above the ladder of which the size
could not be determined PCR of the gpD gene did give the expected band. The PCR product of the gpD gene
was sent for sequencing and matched the sequence of phage lambda perfectly. Based on this it was assumed
that we were working with phage lambda. Still, the new results, combined with a BLAST search of the gpD
gene, which indicated that some E. coli strains carry the gpD gene on their genome, prompted a
renewed
investigation.
To confirm that the phage used was indeed phage lambda, another PCR was performed. In this PCR the phage
stock was directly used as template. As a positive control the phage lambda genome was used, while E.
coli strains DH10B and LE392 were used as negative control. The gpD gene, the PL region and PR+PRM
region were targeted by the used primers. The PCR was carried out using OneTaq and Q5 DNA polymerases.
For OneTaq the PCR was performed with and without DMSO, Q5 was used with DMSO only. The agarose gel,
which is shown in figure 5, gave no bands if the phage was used as template. E. coli DH10B showed clear
bands for gpD in the reaction with added DMSO. E. coli LE392, showed vague bands for gpD with DMSO. The
phage lambda genome positive control shows expected bands. This result shows that there likely is a copy
of gpD on the E. coli genome, while the PRM and PL regions are missing from the E. coli genome. The
clear and correctly sized bands for E. coli DH10B show that the gpD gene is likely present on the E.
coli genome and cast doubt on the validity of the earlier sequencing results. The presence of the gpD
gene also explains earlier trouble during cloning of the spacers, which suffered from very low cloning
efficiency. Self-targeting of the spacers could have caused this.
As the PCR removed our evidence for phage lambda, we looked critically at the TEM image we produced of
our phage stock and compared it to TEM images of other E. coli bacteriophages. From TEM images we
conclude that the phage in our stock is likely to be phage T4 or a closely related phage. Many characteristics of the phage pictured in figure 6 point towards a T4-like phage and not phage
lambda. The head of phage T4 is rectangular in shape, while the phage lambda head is round. The tail of
phage T4 is relatively short compared to the head, while phage lambda has a long tail. At the T4 tail
has a plate at the end, while the phage lambda tail ends pointed. Measurements of the dimension confirm
that the phage is like T4. Average width and length of the head of imaged phages are 75 nM and 103 nM,
in comparison to 86 nM and 120 nM for phage T4 [5]. This is somewhat smaller, which could be caused by the
measurement method and damage to the phages, as the variation between imaged phages is significant. T4 contamination can also explain earlier results. After genome isolation from the phage stock digestion
using restriction enzymes was attempted. This did not result in restriction fragments that could be
visualized on gel. The assumption was made that there was something wrong with the phage genome
isolation method. However, phage T4 modifies its genome through methylation and glucosylation, which
protects it against endonucleases and explains the lack or digestion [6]. Unfortunately, lack of time and lack of phage lambda meant that we were not able to complete testing of
the lambda engineering method and the hybrid phage method. As such, it is not possible to draw
conclusions on the viability of the methods for producing chitin binding phages. Another attempt can be
made, using verified phage lambda, when there is time.
Strains and vectors Phages Phage lambda, more on the content of this in the results and discussion, and phage T7 were taken
from a phage stock created by Marnix Vlot and stored at 4 °C in our lab. Protocols Protocols for genome isolation, Q5 PCR, Gibson assembly of constructs and transformation can be
found here.
PCR verification of phage lambda was performed using a protocol which is suitable to both Q5 and
OneTaq polymerase. Reactions with DMSO contained 0.7 µL DMSO on 25 µL reactions. From the phage
stock 5 µL was used as template. From the phage lambda genome 10 ng was used as template. For
DH10B and LE392, 5 uL of MQ water with dissolved culture was used as a template. For the gpD
primers with phage stock as template a temperature gradient PCR (from 61 °C to 62 °C) was
performed, to give the maximum chance of successfully repeating the earlier positive result. For
all PCRs the following protocol was used: Alternative Phage Engineering
Introduction
Approach
Recombination System
Part
Biobrick Code
Lambda gpD - Chitinase A1 recombination module
BBa_K3286277
Lambda gpD - GFP recombination module
BBa_K3286278
Lambda gpD - HAD recombination module
BBa_K3286279
Lambda gpD - PD1764 recombination module
BBa_K3286280
Counterselection Method
Hybrid Phage
Part
Biobrick Code
Lambda gpD - Chitinase A1 Fusion
BBa_K3286269
Lambda gpD - GFP fusion
BBa_K3286270
Lambda gpD - HAD fusion
BBa_K3286271
Lambda gpD - PD1764 fusion
BBa_K3286272
Phage T7 P10 - Chitinase A1 fusion
BBa_K3286273
Phage T7 P10 - GFP fusion
BBa_K3286274
Phage T7 P10 - HAD fusion
BBa_K3286275
Phage T7 P10 - PD1764 fusion
BBa_K3286276
Protein model
Results
Conclusion
Temperature (°C)
Time
Cycle
98 °C
5 minutes
98 °C
5 minutes
35 cycles
Primer specific Tm
30 seconds
35 cycles
72 °C
30 seconds
35 cycles
72 °C
2 min