In our initial conceptualization of GSHield, we only considered the glutathione production at the boundary between the patch and the oral cavity. However, during our meeting with Dr. Stephen Sonis, he explained that damage from oral mucositis usually occurs from the bottom up. Therefore, sequestering ROS at the level of basal epithelial cells would be more effective than a surface-level response.
Thus, this posed a new question:
Can the glutathione, produced at the surface level, penetrate deep enough into the oral epithelium to affect the basal lamina?
Tight junctions in the oral epithelium break down during cancer therapy, so the penetrability of substances from surface improves over time. Therefore, the idea that glutathione could diffuse from surface to the basal level is possible. However, we were not certain if this would happen.
In order to improve on our previous plan, we decided to develop a model to study glutathione diffusion through compromised epithelium junctions and determine whether we can achieve therapeutic levels of glutathione in deeper parts of the tissue.
Our model was based on a linear 1D nutrient diffusion in cerebral organoid constructs detailed in a paper by Dr. Richard McMurtrey in 2016. Below represents our calculations to produce the model equation:
We considered the following factors when defining our model:
More information on our process to define these factors can be found below:
For some of the above factors such as diffusion coefficient, we had to rely on average/typical values. Other factors that we would like to incorporate in the future include age, sex, body temperature, chemotherapy history (# of cycles, dose, etc), and lifestyle (ex. smoker). Many of these factors would improve our estimate for the diffusion coefficient, thus improving the model; however, we were unable to find suitable data at this time.
Running this model produced a plot of the GSH concentration as a function of depth at a given time. The depths of the stratum basale at different locations in the mouth are marked on the plot. These values were taken from data from Measurement of Oral Epithelial Thickness by Optical Coherence Tomography (Stasio et. al 2019), which measured the thickness of the oral epithelium in non-cancerous subjects. This would have been improved by using data from patients undergoing cancer therapy; however, we found several sources that appeared to have conflicting data and opted to stick with our original source for the time being.
Below is the MATLAB code and snapshot of the plot at t = 15 min, for reference.
Therefore, we now have a way to determine how deep the glutathione will travel, in lieu of experiments on in vivo models. However, this model can continued to be iterated upon.
Our mathematical model was able to further inform the design of the following experiment to help us further improve the project.
Additionally we can use the following experiments to improve upon our model, which can then further inform more experiments.
Our model will also allow us to specify physical design features of the patch, such as the size, balancing a comfortable fit with providing enough surface area of yeast cells to produce enough glutathione.
Our original goal was to design plasmids to import the acsABCD cellulose synthesis operon into E. coli. We decided to follow the design of the 2014 Imperial College iGEM team; thus, we fused acsAB together in one plasmid and acsC and D with a ribosomal binding site (RBS) in the middle of the two genes. However, IDT had synthesis issues, so we couldn’t order the entire acsC-rbs-acsD part as a single gBlock.
We knew that Golden Gate could assemble up to 10 parts reliably, and ordering multiple gBlocks would allow for faster shipping. So, we designed acsC and D as an assembly of 5 smaller fragments.
From here, we followed our existing Golden Gate assembly protocol to build up the plasmid. Next, we tested our plasmid to see if it worked (by running a restriction digest and sequencing). However, we noticed that we kept getting weird assemblies.
To move forward and learn from these errors, we consulted our mentors. While reviewing the assembly in Benchling, we found out that one of the sticky ends was too similar. Thus, it could have been assembling in the incorrect order or in an infinite loop, decreasing assembly efficiency. From this, we redesigned our primers to fix these issues with the gBlock.
McMurtrey R. J. (2016). Analytic Models of Oxygen and Nutrient Diffusion, Metabolism Dynamics, and Architecture Optimization in Three-Dimensional Tissue Constructs with Applications and Insights in Cerebral Organoids. Tissue engineering. Part C, Methods, 22(3), 221–249. https://doi.org/10.1089/ten.TEC.2015.0375
Stasio, D. D., Lauritano, D., Iquebal, H., Romano, A., Gentile, E., & Lucchese, A. (2019). Measurement of Oral Epithelial Thickness by Optical Coherence Tomography. Diagnostics (Basel, Switzerland), 9(3), 90. https://doi.org/10.3390/diagnostics9030090