The main goal of our whole engineering process was to modify the M13 bacteriophage, isolate it and depose it on a gold-coated silicon wafer. This would enable the production of the patterns necessary for our Volatile Organic Compounds (VOC) to bind, leading to colorimetric output.

Specifically, we aimed to modify the phage’s pVIII or Major Coat Protein, which is expressed in approximately 2700 copies in the phage’s capsid. It is a 73 amino acid-long protein, although the first 23 amino acids are cleaved before finalization of the phage particle’s assembly, functioning as a signal peptide, leaving a 50 amino acid-long helical protein on the phages surface, with the N-terminal exposed on the surface. This outer protein region is the target of our modification, where we aimed to insert a short amino acid sequence exhibiting high binding affinity for our compounds. Thus, we first needed to discover these sequences.

Initially, phage display was considered for this task. However, limited laboratory experience, cost and time constraints prevented us from performing it, and we relied on in silico methods instead. More details on this can be found in our first engineering cycle as well as in the Dry Lab category of our wiki.

Once the desired sequences were found, we attempted to integrate them into the phage’s genome using the M13KE vector. For this purpose, we needed to explore our options. Initially, homologous recombination was considered, as it is a relatively easy method for phage genetic engineering. However, only by using this method only a limited amount of phages would potentially obtain the desired sequence, leading to a non-optimized (low engineered to non-engineered phage ratio) sensor [2]. Thus we were able to identify other methods involving mutagenic PCR and traditional cloning that are discussed both in the current page and in our Design page.

During the design phase, bioinformatic analyses were being conducted in order to determine the optimal amino acid sequences to be inserted into the M13 pVIII N-terminal region. For this purpose, a very large number of sequences were being generated and tested sequentially using Molecular Docking (see Model).

Wet lab experimental setup required the presence of an amino acid in position 1 whose codon starts with Guanine: Alanine, Valine, Aspartate, Glutamate or Glycine. This was to incorporate a PstI recognition sequence in the forward PCR primer, and enable the removal of the original sequence later down the line [1] (see Design). The sequence would also need to satisfy sensitivity (high binding affinity) and specificity (lower binding affinity for the other ligands) related conditions. A sequence satisfying the above was, at that point, DYYAW, chosen to bind with Hippuric Acid. Its coding sequence was incorporated into a PCR primer and prepared for ordering.

Forward: 5′-ATATATCTGCAGACTATTATGCGTGGCCCGCAAAAGCGGCCTTTAACTCCC-3′

Reverse: 5′-CACCCTCTGCAGCGAAAGACAGC-3′

Insert sequence in bold, PstI sequence underlined.

Before that, we decided to validate its efficacy in a manner that would more closely emulate real life conditions.

We inserted our chosen sequence (DYYAW) into the pVIII protein sequence at the appropriate position (position 2-6, replacing EGDD) and folded it using AlphaFold2. This would, in our opinion, constitute a better representation of the actual system. We prepared our protein and our ligand molecule appropriately and performed molecular docking simulations using Autodock Vina software.

The folded modified pVIII protein produced a much worse result than initially anticipated, with binding affinity scores dropping from -4.407 kcal/mol to -3.398 kcal/mol. This constituted an unacceptable drop in sensitivity for our potential future biosensor.

We attributed this disappointing result to the presence of Alanine in position 1, right before the point of insertion. To validate our results, we also performed docking with the hexapeptide ADYYAW and found it to closely match the folded protein (-3.443).

We concluded that we should continue testing hexapeptides, starting with Alanine and followed by a five amino-acid sequence satisfying the other conditions stated in the “Build” part of the cycle. Introducing a fixed Alanine would not affect the number of peptides we had to test, and would only minimally increase docking time for each receptor since the average search grid needed would be slightly greater.

Once the Dry lab concluded to one optimal sequence for each VOC, we had already designed the method to insert it in the M13 phage. In order to genetically modify the bacteriophage we used M13KE vector DNA

According to dry lab results M13KE DNA needed to be modified in order to incorporate the following genetic modifications in the gene of pVIII protein:

Literature research, provided us with PCR based protocols for M13KE phage engineering [1], which depended firstly on site-directed mutagenesis PCR, in order to remove the unique PstI restriction enzyme recognition site of the vector, and then on a second mutagenesis PCR in order to induce the desired modification in the pVIII gene. More information about the selection and the design of these PCRs can be found in our Design page.

The strategy is depicted in brief, graphically in Figure 1.

To obtain the mutagenic M13 phage, two PCR reactions were performed as described above. To verify their efficiency, 5μl samples of each reaction were evaluated, by loading them on a 0.8% agarose gel (Figure 2). The DNA band of the linearized M13/ the product of the PCR, was expected to be seen at ~7200 bp and thus between the second and third ladder band. However, no such bands were observed and we came to the conclusion that the reaction was not successful.

After our initial results several repetitions of the PCR reactions were performed, examined by electrophoresis. These attempts were preceded by constant evaluation, consultations with our advisors and our own research. During our troubleshooting various thermocycling conditions and reagent’s quantities and ratios were examined with no successful results. (examples can be seen in Figure 3). In brief, we altered the number of cycles, annealing and extension time and temperature, initial and normal denaturation time and reaction volume. We also tried to linearize the M13 genome before the reaction, add DMSO, preheat the primers and change the thermocycler we used.

The nature of our results, and more specifically the absence of clear bands, even of non-specific products, and the high frequency of smearing, indicated that the most probable cause of failure lies in the design of our primers given that we have performed extensive troubleshooting to pinpoint other potential causes [3]. After further investigation, we discovered that ΔGs’ absolute value of the formation of hetero-dimers was higher than values recommended by many primer designing tools. Thus, new codon choices of the mutagenic primers were made in order to minimize hairpin, homo and hetero dimer formation, using the IDT OligoAnalyzer tool. In order to avoid wasting time, we made three different pairs of primers with slightly different sequences, length and GC content, to simultaneously test all three in order to theoretically have more chances for success. We designed partially complementary mutagenesis primers, with 1-3 non-complementary bases on their ends, to minimize dimer formation and improve the chances of primer-to-template binding. All the newly designed primers had a lower tendency to form hetero-dimers and can be seen in “Design”. To build upon the new design, further PCR reactions were attempted with the new sets of primers. Again many different cycling and mixing conditions were used. To verify reactions efficiency, 5μl samples of each reaction were evaluated, by loading them on a 0.8% agarose gel (Figure 4).

Since the desired phage was not obtained with this technique, the “Test” step could not be performed in this iteration of the cycle.

Unfortunately, we still did not observe the desired product. We concluded that the specific PCR reaction to induce the desired mutation wasn’t as effective as we had initially anticipated. Thus without further ado, we agreed upon redesigning fundamentally. We changed our strategy to traditional cloning technique and discontinued the use of PCR mutagenesis to obtain our modified phage.

We considered inserting the desired sequence in the pVIII protein gene, by designing and ordering gBlocks/Gene Fragments. Specifically, the sequence would be incorporated in a larger insert fragment. This fragment, which has a different restriction enzyme site on each of its ends, corresponds to the original fragment of M13KE sequence, with the difference of a set of 15bp, that represents the desired modification.

We decided that the insert sequence would be cloned into the M13KE vector with double restriction, using KpnI and BsrGI enzymes. These fragments would have a length of 593bp.

In order to distinguish modified from unmodified M13KE DNA, a new HindIII restriction site was introduced, by inducing a silent mutation inside “M13 geneIII”. When digested with HindIII, the genetically engineered phage would produce two fragments (4649 bp and 2573 bp), instead of the one product of 7222 bp of the unmodified M13KE produced by single-site digestion.

The green arrow represents the designed gBlock/Gene Fragment sequence.

To acquire the wanted construct, we performed double digestions as described above, on both the M13KE vector and the insert. The desired digested products were visualized through loading on an agarose gel (outlined on Figure 5), and then extracted using a “Gel Extraction kit”. Through electrophoresis of the extracted product, clear bands were seen at ~600 bp and ~4500 bp respectively as expected (Figure 6).

In order to test if the cloning was successful, we proceeded to chemical transformation of XL1-blue cells with the ligation product. A positive (PUC19 plasmid) and negative (only cells) transformation control were added, as well as a ligation negative control (only digested vector, no insert). Some results of the first few attempts were the following:

Figure 7: All cultures showed phage growth, including the negative control. After examining the DNA sequences of a few plaques, by isolating it and digesting it with HindIII, we concluded that the plaques were the result of contamination from unmodified M13KE.

Figure 8: Only the positive control showed phage growth, meaning that the ligase reaction was probably unsuccessful.

In this iteration of cloning, it was imperative to rebuild by testing various different vector:insert ratios in the ligation process (for more detailed information check “Lab Notebook”).

Figure 9: The growth was as expected at positive and negative controls. Eicosane and hippuric 1 sequences, plaques were also observed. This indicated that the engineered phage might be finally constructed.

From the results shown above, plaques - and thus hopefully modified phages - had occurred. The corresponding ligation reactions were the only ones where the molar vector:insert ratios were changed in favor of lower insert quantity, (to 2:1 and 3:1 respectively) showing that the ratios could be the source of the problem for our previously failed experiments.

Figure 10: In this attempt, we used the molar ratio that gave us plaques in all of the ligase reactions and got plaques from all the sequences tested.

From the results shown above, plaques - and thus hopefully modified phages - had occurred. The corresponding ligation reactions were the only ones where the molar vector:insert ratios were changed in favor of lower insert quantity, (to 2:1 and 3:1 respectively) showing that the ratios could be the source of the problem for our previously failed experiments. To ensure that the plaques were indeed formed from modified phages, we isolated DNA from a few of them and proceeded to digestion with HindIII.

Before digestion with HindIII samples were loaded in 1% agarose gel producing the following results:

Most samples show three bands: 1) between 2kb and 3kb, 2)between 3kb and 4kb 3)between 6kb and 8kb corresponding to the circular M13 DNA.

Results after digestion with HindIII (the incubation time was induced due to lack of time)

Most samples showed two to three zones: 1) between 2kb and 3kb 2) between 6kb and 8kb 3) at 4.5kb. Samples 4 and 13 show another unique band between 2kb and 3kb that could potentially correspond to one of the two fragments expected to result from HindIII digestion in two sites since two HindIII sites exist only in the modified bacteriophage.

Modified M13 DNA is expected to produce 2 bands at 4600 kb and 2600 kb when digested by HindIII. These bands are observed in the current electrophoresis in samples 4 and 13. It should be noted that the two bands did not coincide with the ones before HindIII digestion.

Therefore, highlighted bands serve as an indication of the possibility of having successfully modified the bacteriophage.

From the above, we conclude that we have evidence of having genetically engineered M13 bacteriophages modified to bind with Hippuric Acid and Perillaldehyde.

In order to repeat and verify the results, an isolation-digestion procedure needs to be performed again but with longer incubation time, in order for the enzyme to produce more of the digested product. The electrophoresis will also last for a larger period of time, running at lower Voltage, in order to be able to have clearer bands. In parallel, the rest of the samples 4 and 13, will be sended for sequencing.

Restriction digestion with HindIII suggests that there is high probability that the genetically engineered M13 bacteriophage for the binding of Hippuric Acid and Perillaldehyde have been successfully constructed. Given that the success occurred in ligation mixtures with high vector/insert ratio we conclude that there is a strong probability of this ratio to be a significant factor for the success or failure of the ligation.

Despite the construction of our M13 bacteriophage, our extremely limited time was prohibitive for further repeating the procedure for verification, amplification, purification and then testing of the modified M13 bacteriophage.

For the further development of our project, we would test the binding of the bacteriophage using Tryptophan Fluorescence Assay as described in our “Design” page. Results of the binding assay would be very informative for evaluating the effectivity of our molecular docking simulations and could serve as a guide for further optimizing amino acid selection, not only for the detection of the specific VOCs, but also for other compounds associated with diseases such as cancer [4].

Equally imperative to our project, is the test of the colorimetric output of the engineered bacteriophages when self-assembled on gold-substrates, which constitutes our end product. This test would serve as an evaluation of the effectiveness of the engineered M13 bacteriophage as a colorimetric-biosensor.

Nevertheless, despite the indication of having successfully engineered two types of M13 bacteriophages (for Hippuric Acid and Perillaldehyde binding) it should be noted that the modifications probably affected their viability. This can be derived by the fact that single plaques of the modified phages were observed with no prior dilution, something that is not common for M13 bacteriophage [5]. This suggests probable alternation of the viability of the bacteriophage, which could also serve as an explanation for failed transformations resulting from non-reproducible bacteriophage particles.

In order to explore the potential scenario of M13 phage particles losing their infectivity by the genetic modification of pVIII protein, we would need to redesign our strategy for choosing amino acid sequences that bind to the VOCs. Specifically, we would take into consideration physical and chemical alterations that could probably significantly interfere with the bacteriophage’s life cycle. For example, the electrical charge of the N-terminus of the pVIII protein could reduce the viability and/or infectivity of the particles [6]. That means that during a future bioinformatic analysis the charge of the amino acid sequence needs to be taken into consideration.

In conclusion, our engineering involved multiple iterations of the design-build-test-learn cycles, recruiting both wet and dry lab methods. Every engineering cycle served as a stepping stone for the development of the genetically engineered bacteriophages. The last engineering cycle ended with promising results of producing the engineered M13 phages designed to bind with Hippuric Acid and Perillic Aldehyde.