Team:OhioState/Engineering
Engineering Success
Dry Lab
Kinetics Route
Once we had decided on our proteins, we wanted to be able to predict which one would work best, before testing them in the lab. The metric for “best” would be the protein that binds the strongest to lipid A, with a high rate. We conducted research to find kinetic data and rate constants for each modifier reacting with lipid A. This data was not available, so we returned to searching for ideas until we found docking.
Docking
Using crystal structures, the docking method outputs docking scores for a protein binding to a substrate. Originally, this seemed perfect, our team had both proteins and a substrate, so we expected to be able to get a docking score. However, after talking with Austin Cool, it became much more complicated than that. Firstly, docking primarily uses Gibbs Free Energy calculations to determine the docking score, but lipid molecules (like lipid A) notoriously sidestep and disrupt these calculations due to their long carbon chains (leading to inaccurate results). Secondly, there is no method to know if the way the docking program orients and folds the proteins is accurate, since the proteins aren’t well documented binding to lipid A. Because of this, even if we got results, we would have no way to validate them and no way to do a rough check to see if the model results even make sense. We would just have to blindly accept them. Thirdly, with only seven proteins, it would be relatively easy to get experimental results and discover the best anti-lipid A protein to use. This would eliminate all the guesswork inherent to our model, effectively making the docking model not worthwhile.
Mutation Model
It wasn’t until meeting with Dr. Strathdee and Dr. Patterson that we turned our attention to a bacterial mutation model. Bacteria mutating away from the select phage is a critical issue with phage therapy. We thought we might try to create a model to give a rough estimate of the amount of time it would take for the bacteria to mutate in such a way that renders our lambda phage non-functional. After another round of reaching out to experts and consults, we had to alter this idea. The randomness of mutation made it hard to give a set time as to when the bacteria would mutate out, it could be hours or it could never happen at all, it was all random and equally likely.
Bacteria-Phage Population Model
Our final model iteration came when we talked with Dr. Raman and adapted a bacteria mutation model to a bacteria-phage population model. Given initial parameters, we found steady state concentrations of both bacteria and phage. We initially used parameters unspecific to E coli. and Lambda. Once we had the process set, we changed these parameters to reflect both our specific bacteria and phage. In doing so, we had outputs closer to what the actual E coli. Lambda Populations would look like.
Wet Lab
Anti-lipid A molecules
Once our team researched and decided on our project, using phage therapy to target lipid A, the first part of the project we needed to solve was what molecule can we use to target lipid A. After in depth research, we compiled a list of over 20 different molecules that we thought could produce the results we wanted. These molecules varied widely in how they interacted with lipid A and what host species in which they were naturally found. We then took our list of molecules and met with multiple researchers who study sepsis, including Dr. Steve Abedon, Dr. John Gunn. Taking what we learned from them, we whittled our list down from 20 molecules to seven individual molecules and one pathway of molecules. In addition, we researched various assays that would actually indicate if we had lipid A in our final product. However, after consulting with these researchers, we dropped most of our tissue culture assays in favor of a HEK TLR4 reporter cell line.
FraR
Another important detail in our project was our choice of promoter. The problem we had was that the genes in our phage are toxic to E. Coli, so we needed our genes products to not be present in the lab strain E. Coli, in which the phage grew, but present in wild type E. Coli. This problem was solved by implementing a FraR-pFraB system. The FraB gene is involved in fructose asparagine metabolism and is strongly repressed by the FraR protein when not in the presence of fructose asparagine. The E. Coli we would use to develop our phage would not express our anti-lipid A molecules, but wild type E. Coli lack this gene and would express it. Therefore, our system would be active as a promoter in wild type E. Coli and completely inactive in the lab strains.
We designed the experiment initially to test whether our promoter would function correctly or not by creating reporter strains by cloning our promoter into a Lux reporter pSB401. Then we transformed into a strain that expressed fraR inducible by IPTG via a plasmid. We then attempted to transform into a strain that didn’t express fraR. However, upon initial testing we realized that strain that contained two plasmids with two different antibiotic resistances would grow slower and thus have lower luciferase levels. So we transformed the strains without the fraR plasmid with an empty vector containing the antibiotic in order to ‘burden’ both strains a similar amount. After doing this we observed that pFraB worked as expected with strong repression in the fraR(+) strains with IPTG.
However, we were wondering whether we could create a promoter with the same regulation as pFraB but a stronger expression. This was because we want our phage to cause the expression of these genes very fast and at a high level before the cells lyse from the immune response. So we decided to add the pR promoter(BBa_R0051) upstream of the pFraB promoter(BBa_K3783000) to create a combination promoter "pC" (BBa_K3783001). So we could create a system that would highly express in our wild type E.coli but be repressed in our lab strain so we could grow our phage clones. When we went to test our promoter we saw a 30 fold increase in promoter strength. This promoter also maintained the regulatory properties of pFraB with strong repression from fraR. The protein fraR was induced by the addition of IPTG as seen previously in the fraR(+) strains.
