In order to simulate the two-cell consortium model, we used Simbiology to model our system with the hopes of testing the sensing ability and logic-gate functionality of the system (see Model). In order to determine reaction rates and product concentrations over time, we modeled a number of different reactions using various kinetic equations within Simbiology. Each of these reactions required the related association, dissociation and kinetic constants in order to create an appropriate model. While our ability to determine these constants was limited initially, by surveying literature and making frequent improvements to our analytical procedure, we were able to create a computational pipeline that could predict our binding constants.
We initially planned on finding these constants using previous literature and making reasonable constant approximations based on proteins of similar function. This is a strategy that has been used by numerous iGem teams in the past, and we therefore believed it was appropriate. However, after reviewing the literature related to the HrpS, HrpV and HrpR proteins, we found there were little to no studies that explored their binding and activity to their respective ligands and proteins. Further complicating our issue was the fact that HrpR and HrpS, which we assumed to be the rate limiting step for this set of proteins, created a dimerized complex before acting on its ligand. This made it difficult to approximate constants based on proteins of similar function. The lack of information was a huge issue, as this set of proteins was crucial to the logic-gate functionality of our system, and without appropriate constants our model’s precision and validity were limited.
To combat this issue, we surveyed literature for tools that could determine the molecular dynamics of proteins based on their 3D structure. We decided to use transcomp, which can predict the Kon and Ka (from which we could calculate Koff) for bound proteins. In order to determine the structural binding of dimerized proteins, we decided to use Hawkdock, a well-cited method that uses a docking algorithm to determine protein-protein interactions. A more detailed breakdown of this pipeline can be found in 2.3.2 of our Modelling section.
Despite creating this pipeline, we were still missing the PDB files of the proteins in the HRP system that described the 3D structure of each protein. These files were necessary for Hawkdock to determine dimer interactions, and for transcomp to determine protein binding. While we surveyed the literature and relevant protein and gene databases, we could not find the PDB structures for the proteins in question, and could only find their amino acid sequences. At this point, we were left with an unfinished pipeline.
However, the July 2021 release of AlphaFold by Deep Mind gave us our final piece of the puzzle. Alpha Fold is a revolutionary tool that can predict the 3D structure of a protein given its 2D amino acid sequence. By using AlphaFold we could generate the necessary PDB files, and create predictions for our dimerization reaction kinetics using the aforementioned tools. From this our Dry Lab Team was able to design a streamlined procedure that could predict the association, dissociation and affinity constants for dimerization reactions only from their amino acid sequence. By consistently determining the limitations of our procedure and seeking new tools from the literature, we were able to design a computational pipeline using novel tools that can be used to find constants when literature is limited, a problem that is frequent for iGem Dry Lab teams.
Due to limitations in our access to the wet lab, we were unable to move past certain roadblocks in the building phase of our experimentation. However, we had designed a series of procedures that involved testing and troubleshooting at different expected checkpoints. This included testing and troubleshooting all of our system components individually to ensure their independent success. Then, we would build different combinations of our components to see their success in pairs before moving on to assembling the entire system.
Our Wet-Lab engineering design process began with the actual designing of the gene fragments, which took quite a bit of trial and error in itself. First, we gathered the necessary gene fragments for the different components of each receptor system from various sources. Once we inputted these into the Benchling software, we worked on combining the individual parts (e.x promoter, target protein, restriction enzyme sites, RBS, and fluorescent marker). Then, we designed primers for each receptor system. This was done through designing and redesigning to get to the optimal sequence of base-pairs, to be checked on by our PhD advisors. Once the primers were complete, we would input our sequences into IDT to scan them for complexity, and more often than not we were faced with over-complex sequences that needed to be optimized. Thus, we would go back to the drawing board and fix up the sequences in the software then proceed to re-input them into IDT. Finally, once we had IDT-approved sequences, we ordered them.
Since our system is quite complex with multiple sub-parts playing into the overall design, we decided it would be best to test the individual sub-systems individually, then gradually move into testing them together, and then finally the whole system as one. While we were unfortunately not able to carry through our planned series of experiments in-lab due to time and resource constraints, we did have a thorough step-by-step experimental system to ensure the success of each sub-part before moving on to the next. This set of experiments can be found on our Experiments page, but involves measuring the success of each sub-part before testing the logic-gate functionality of our system.
We were hoping that by theoretically conducting these experiments in a methodical manner, we would be able to establish clear checkpoints for troubleshooting. For example, while initially working with the indole system, we faced some trouble. Since we had to split the indole system gene sequence into two fragments, we had to perform PCR assembly to assemble them together. However, we faced quite a bit of trouble with this as this was a procedure none of our members had performed prior and we had very minimal guidance. As such, we decided to continue researching and trying to troubleshoot the proble with the indole sequence while simultaneously beginning to work with the butyrate and gaba sequences as we were able to get both as full single sequences. This was an important part of our design process, because it allowed us to recognize the issue we were facing in the building and testing phase of the design cycle, work towards learning what went wrong and how we can move forward, while simultaneously testing other systems that we knew we would not face the same issues with.
Our trouble shooting steps typically went as follows: