Partnership with the TU Delft iGEM team
One of the important contributors to our project was the TU Delft iGEM team. Through our partnership, we were able to support and improve each other’s projects by considering different modeling perspectives for T7 bacteriophage: population dynamics and gene products. We were able to better understand how to maximize signal production and how we can alter our product to make it more effective in the future.
Our Respective Projects
TU Delft iGEM team project: The TU Delft iGEM team is working on genetically modifying T7 bacteriophage to produce a toxin, Cry7Ca from the organism Bacillus thuringiensis, in the gut of a locust to help combat the locust crisis. In the wet lab, they are focused on replacing three non-essential genes in the T7 bacteriophage, 0.6, 1.1, and 4.3, with a GFP gene as a proof of concept before testing the engineered genome with the toxin. In addition to the wet lab experiments, the TU Delft team also aims to understand phage-host interactions in a modified Beretta and Kuang model. This models the amount of susceptible bacteria (S), amount of infected bacteria (I) and the amount of phages (P) to calculate values like number of cells lysed (Beretta and Kuang 1998). The TU Delft iGEM team added in an equation that describes the amount of GFP produced over a period of time and how substrate availability affects the phage and GFP production.
Our Project: Our iGEM team is working on creating a T7 bacteriophage that is genetically modified to produce a signal when it lyses E. coli. We modeled with the T7 Bacteriophage with the modifications we hypothesized would produce the most signal, using a software, PineTree. Pinetree models the infection cycle for one phage and outputs the gene expression. With the outputs of PineTree, we created a burst size and lysis time calculator. Using Pinetree and the burst size and lysis time calculator, we attempted to find the best possible genome to produce the most reporter protein.
Shared Project Goals: Both of our teams are attempting to re-engineer T7 bacteriophage and are specifically focusing on creating a larger reporter signal. Furthermore, both of our groups are using GFP as a proof of concept to visualize the effects of modifying the T7 genome. However, each of our teams had a unique approach to modelling the effects of engineering the T7 bacteriophage genomes. The TU Delft team focused on modeling phage-host interactions, while we modeled the effects of mutating the genome on the infection cycle of the T7 bacteriophage. This made the TU Delft iGEM team the ideal, mutually beneficial partnership.
Meetings with the TU Delft iGEM Team
Throughout the duration of this project, we were able to meet up with the TU Delft iGEM team multiple times virtually with Zoom calls.
During the first meeting with the TU Delft team, we were able to understand each other’s goals and perspectives coming into this project. We also explained our individual approaches to achieve the shared goal of maximizing reporter protein production. We particularly focused on understanding the basis of the models each team was utilizing to maximize this output. Through this meeting, we were able to gain an introduction to the model that TU Delft iGEM team was using to model phage-host interactions. This was very different to the Pinetree model that we were using, which focused on the output of protein from individual phages. Because we had different approaches to reach a similar goal, we decided that we could create a mutually beneficial partnership. After this meeting, we decided to send more specific details and background information about our models through email.
After reading more about the basis of the TU Delft model, we decided to have another meeting with the TU Delft Team. During this meeting, we were able to delve deeper into the specifications of each model. We discussed the assumptions of each model and brainstormed hypothetical simulations we wanted to run. With this meeting, we were able to take this broader concept of phage host interactions and relate it back to our project. With the TU Delft model, we would be able to see which genome would be able to lyse the most cells and be most effective from a population dynamics standpoint. From our model, the TU Delft team would be able to simulate how gene replacements they were attempting in the wet lab would potentially look like, and also predict how inserting the toxin, which has a longer length than the GFP gene that was used as the proof of concept, in the T7 genome would affect lysis time and burst size. After clearly understanding the goals of our teams, we decided to work on sending each other the specific information needed to run each other’s simulations.
Results of Our Partnership
To help the TU Delft iGEM team, we were able to run a few simulations through our PineTree system. Using our Breseq software, we were able to model the replacement of the genes 1.1 and 4.3 with an eGFP gene that the team planned to use in the wetlab and also model the toxin the TU Delft team hoped to test and include in their final product. Furthermore, they wanted to compare the native toxin to the codon optimized toxin to understand which one would be more beneficial to their project.In total, we ran eight simulations for them, replacing the genes 1.1 and 4.3 with eGFP, CryToxin, etc. For each simulation we created three graphs, showing the average expression of the protein, the lysis distribution, and the burst size distribution. This allowed them to determine the effect of their insertion on the phage fitness. With our simulations, they were able to determine that replacing the 4.3 gene with an eGFP gene would be better than replacing gene 1.1. Furthermore, we suggested inserting the toxin or reporter signal in a more expressive area, such as later genes to increase transcription, such as 10A.
For us, this partnership helped us understand the T7 bacteriophage by examining population dynamics. Our PineTree model helps us simulate the infection of an E. coli cell by the T7 bacteriophage. For each mutation, we are able to run 96 simulations and aggregate and analyze this data through our R scripts. The TU Delft iGEM team is taking into account population dynamics with the mathematical model described in the Beretta and Kuang paper (1998). Through their model, we can assess the interaction between phage and susceptible bacteria. By taking the lysis time and burst size per mutated genome (Table 1), we calculated from our own model, along with varying bacterial cell densities we aimed to test (Table 2), the TU Delft team was able to calculate the amount of cells lysed per 100mL. The TU Delft team made the following assumptions: phage death rate = 2 phage/day, bacterial contact rate constant = 6.7e-8 ml/day bacterial growth rate = 1.34 bacteria/day, carrying capacity of marine environment = 2e6 bacteria/ml.
Table 1: The three sets of initial conditions provided to the TU Delft iGEM team to run through their simulation.
|Genome||Lysis Time(s)||Burst Size|
Genomes were selected because they decreased lysis time, maximized burst size, or maximized GFP production to understand the effect on number of cells lysed.
Table 2: The E. coli Cell Densities that were tested.
|E. coli Densities|
Values were selected by the maximum allowed E. coli cell densities in the military: 1 CFU/100mL (SBIR 2017) and recreational water areas: 88 CFU/100mL, 126 CFU/100mL (EAI labs; EPA 2015).
We provided the values shown in Tables 1 and 2 to the TU Delft iGEM team. With their modified Beretta and Kuang model, the TU Delft iGEM team was able to input the lysis time and burst sizes for the genomes we wanted to test to create a set of graphs. In each graph, there were four measurements that were calculated: X_s = susceptible cells (number/ml), X_i = infected cells (number/mL), X_lysed = lysed cells (number/mL), P = Phages (number/mL). For our project, the measurement that is relevant is the X_lysed. This measurement is on the bottom right of Figure 2 and Figure 3. In these simulations, if a lot of cells are able to lyse, we can be fairly confident that we are able to detect the E. coli concentrations listed in Table 2.
With their simulations, the TU Delft team found that the highest number of cells lysed was with the gfp-h-gp10 edited genome detecting 10 CFU/ml of E. coli (Figure 2). In this genome, we were able to see that only about 2 cells lysed per 1mL.(Figure 2, bottom right) This shows us that at the bacterial concentrations we wish to detect, it is unlikely that we would be able to lyse many cells.
To increase the number of cells lysed, the TU Delft team suggested increasing the concentration of bacteria to be able to see more cells lyse. The simulation in Figure 3 was run at a concentration of 1e7 CFU/mL. This graph, with the specifications shown above, showed that about 125 cells lysed per milliliter. This showed us that to detect the small concentrations of bacteria, we needed to implement a filter to create a higher host density environment and lyse more bacteria. This is something we will continue to consider in this project because we need to be able to detect lower concentrations of E. coli that is required by the communities we contacted in Human Practices.
Overall, the partnership with the TU Delft iGEM team was very successful. Both of our teams gained new perspectives on modeling the T7 bacteriophage and helped each other brainstorm different ideas. From this partnership, we were able to understand that the concentrations of bacterial cells we aim to detect is unlikely to have many cells lyse. Therefore, we need a filter system to increase the concentration of the bacteria to lyse more cells. Furthermore, we realized the gfp-holin-gp10 genome would also be able to lyse the most cells. For the TU Delft iGEM team, we were able to provide insight into their wet lab results and offer recommendations about different locations to place their reporter gene in the T7 bacteriophage genome. We are looking forward to using these suggestions and information that we have learned in future advancements for this project.
Beretta, E., & Kuang, Y. (1998). Modeling and analysis of a marine bacteriophage infection. Mathematical Biosciences, 149(1), 57-76. doi:10.1016/s0025-5564(97)10015-3
A Device to Rapidly Detect Coliform Bacteria and Escherichia Coli in Field Water Samples. (2017). Retrieved from https://www.sbir.gov/node/1208355
Lake Sampling for Bacteria. (n.d.). Retrieved from http://www.eai-labs.com/assets/docs/lake_sampling.pdf
Recreational Water Quality Criteria. (2015, October). Retrieved from https://www.epa.gov/sites/production/files/2015-10/documents/rwqc2012.pdf