Our engineering design cycle can be divided into two iterations: one that is based on computational work and simulations and another that involves wet lab work.
Our main goal is to be able to successfully create mutations in the T7 bacteriophage genome computationally, simulate our mutant genomes through Pinetree, and visualize our data using RStudio, to ultimately engineer a bacteriophage with maximized GFP production and minimized lysis time.
Summary of Computational Engineering Design Cycle:
The computational part of our project involved researching topics related to lysis time, burst size, conditions for lysis, gene deletions in the T7 genome, and the logistics of inserting GFP and producing a detectable amount of GFP. Then we designed various versions of the T7 genome with one gene deletion for all the possible gene deletions. We used Pinetree to determine the protein production of these mutant genomes compared to the wild type genome and graphed them using RStudio. We were able to analyze these graphs to determine which edits had the most potential to be combined to create the most efficient and effective phage. After deciding on the edits of interest, we broke into groups for further genome designing, testing, and analyzing for those specific edits. These included edits associated with lysis time, GFP production, RNAse, and the lysozyme mutant. After obtaining data from each of those groups, we were able to combine the most beneficial gene edits, in particular the GFP and lysis mutants, to create possible final genomes, which we worked with TU Delft's iGEM team to test with simulations of our phages' infections entire populations of E. coli (Escherichia coli) rather than just one cell.
The following is a more stepwise description of the "Summary of Computational Engineering Design Cycle" section above:
T7 bacteriophage is a widely studied model organism, so there was plenty of prior research on its genome structure , lysis behavior, burst size, and non-essential genes. We also researched topics such as how much GFP should be produced to be detectable, and what conditions affect lysis (pH, temperature, etc.) in order to get a better baseline for our expected results.
Using the early research that we compiled, our team determined the purpose for our engineered phage: E. coli detection in water. We envisioned a fast-acting colorimetric reporter-based mechanism that would allow users to quickly and accurately assess water samples for E. Coli through the infection of T7 and release of GFP at cell lysis.
Using a systematic approach we designed versions of the T7 bacteriophage genome with each individual gene deleted in order to determine the importance of each gene in regards to lysis time and burst size for the purpose of determining which genes to delete and which genes must be maintained. Additionally, we tested various mutants with GFP and the holin gene inserted into different genomic loci in order to determine the locations of highest expression and fastest lysis within the genome. We developed calculators for lysis time and burst size in order to test our constructs.
To "build" the T7 bacteriophage genomes designed in the previous step in a way that could be tested, we made edits to the T7 bacteriophage genome code using Breseq, an application used to apply mutations to a GenBank file that could be run through Pinetree (click here for more details).
We tested each of the T7 bacteriophage genomes using the Pinetree simulator to determine the effect of each mutant on gene expression trends throughout the genome. Furthermore, we used Pinetree to simulate infections with each of mutants to determine and the amount of GFP produced and holin produced. We were able to collect large samples of data quickly by using the Texas Advanced Computing Center(TACC),to run large numbers of Pinetree simulations of each mutation quickly, rather than running each simulation one at a time on our own computers.
For each of the mutated genomes, we can visualize and compare the wild type and mutant protein production, as well as burst size and lysis time, using a set of R scripts that we designed. Once we obtained the data from Pinetree for each of the mutants, we used these scripts to determine which mutant have the potential to yield optimal lysis time, burst size, and GFP production. The analysis and graphs generated from R helped us determine the maximum GFP expression, lowest lysis time, as well as changes in burst size due to different mutations. Furthermore, we have a better understanding of which genes can be safely deleted as well as the overall expression trends of the T7 genome. We evaluated these results to determine which mutations are most ideal for our goals. Additionally, we now have a better idea for future research and experiments
After researching about the various T7 genes and their functions, and modeling our mutants, we were able to refine our investigations to those specific genes. We used the data that we collected and analyzed in the "Test" and "Learn" steps and used it to determine what areas needed to be further researched, what genomic edits needed to be created and tested, and what areas of code needed to be fixed for bugs and errors, and how Pinetree can be improved. We also determined what parts of the T7 genome were optimal insertion sites for GFP during the mutations we ran, and incorporated these sites into our final model. We also updated our lysis time calculator after initial experiments to determine the protein expression at the lysis time for each individual simulation, rather than using an average lysis time for each simulation and then determining the protein count at that point. We reran some simulations using this updated calculator and ran some simulations for other replacements and insertions. We then began the engineering cycle again, by researching topics for information needed for the next steps of the project to be continued. After going through this cycle with all the mutations we created, we were able to obtain lysis time and burst size graphs for all of our tested mutations and determine what mutated genomes we want to test in wet lab.
In conclusion, our team was able to design an engineered T7 bacteriophage with optimized GFP and lysis time through our design, modeling, and analysis, and we are now ready to assess our construct in lab. This designed T7 bacteriophage can be seen below in Figure 1.
Summary of Wet Lab Engineering Design Cycle:
After determining which genetic edits have the optimal lysis time, we will continue with the engineering design cycle. We'll begin by researching the most effective methods for implementing our genetic edits for T7 bacteriophage. We will then design the new bacteriophage based on the data that we collected from Pinetree and the ideas that we found during the research stage. We will use this design to build the new engineered strain of T7 bacteriophage and test it by determining its effectiveness and sensitivity at determining whether or not E. coli is in a water sample. Once the testing is completed, we will examine the data obtained and find new ways to improve the bacteriophage and procedure to create the ideal bioassay. Specifically, we are using yeast assembly to create our mutated phage genomes and phage titering to measure the amount of phage made in infections. Flow cytometry will be used to measure GFP production of our mutated genomes. What we learn in the computational engineering cycle will determine what mutations we test in the wet lab.