Initial Screen
The primary goal of our project was to create a foundational advancement in synthetic biology by conducting a large-scale screening of several promoters and molecules to identify which pair can be used for biosensor development. Therefore, the first step of the research was to screen each of the promoters against each of the molecules. There are a total of 21 promoter plates, each with 95 promoters and 1 control well. We also selected 10 molecules for the screening: PBA represents 3-Phenoxybenzoic Acid, LOV represent Lovastatin, PRO represents Propoxur, DEP represents Diethyl Phthalate, TAR represents Tartaric Acid, CAR represents Carbaryl, BHL represents Butanoyl-Homoserine Lactone, PGA represents Phenylglyoxylic Acid, PFS represents Perfluorooctane Sulfonate, and CND represents Cis-Naphthalene Dihydrodiol.
In our first set of experiments, we measured the fluorescent signal produced by the molecule and the promoter. Disregarding any wells with an OD600 value less than 0.1 due to a lack of E. coli growth, we selected promoters with a sfGFP/OD600 < 1.5. However, due to a lack of sufficient significant molecule-promoter pairs, the cutoff was lowered to 1.35.
Titrations
Once the top 8-12 promoters were identified for each molecule, titrations were performed to observe the change in fluorescence as the concentration of the molecule changed. An effective biosensor is meant to measure the amount of a molecule present. Therefore, the promoter should produce a stronger fluorescent signal in the presence of greater concentrations of the molecule with a large dynamic range. Concentrations from 0 M to 10 mM were tested, as shown below:
After the titrations, there was a lack of promoters that produced increasing fluorescence proportional to the increasing concentrations of the molecules, despite producing high fluorescence during the initial screening process. Additionally, the initial screen fluorescence sfGFP/OD600 values were much higher than the fluorescence at the same molecule concentration in the titrations.
In order to troubleshoot this issue, we noticed that several of the started up E. coli strains did not produce an OD600 value > 0.1, indicating there was not enough cell density to produce an accurate fluorescent reading. The strains were regrown and measured to ensure a high density before repeating the titration experiments.
While this gave us a few more promising results, there was still a largely linear trend rather than an exponential one with varying concentrations of the molecule. Through further investigation, we found that the LB broth we were using for the E. coli strains was contaminated, which could have affected the results. After the media was remade under sterilized conditions and autoclaved after dilution to remove any contamination, the titration was rerun. The titration values were compared to the initial screening values again.
As shown above, the titration values were slightly higher with the new, uncontaminated media, but they still did not reach the level of the original screen, which is a problem. We decided to test another variable to address this issue: the freshness of the strains. Some of the strains were several days old when they were tested in the titrations, whereas the strains were fresh from the day before when the screening was conducted. We regrew the strains and conducted the titration again after they had been in the incubator for only one day.
The titrations were only repeated with the top 4 strains of each molecule as a sample size of the total number of strains. This conserved time and resources so that more variables could be tested in order to understand the root of the problem. This set of experiments once again improved the increasing trends of the data with increasing concentrations of the molecules and increased the level of fluorescence closer to that of the initial screening, but it was still slightly lower than expected.
Since the titration has been repeated so many times, our team decided to repeat the initial screen with some plates in order to ensure the screen was replicable. Specifically, we used AZ02, AZ03, and AZ11. The new screen (in orange) matched the old screen values (in blue), indicating that the results are reliable since it can be repeated.
We also broke down the data by compiling the fluorescence levels of a single promoter with each of the 10 molecules. Since it is highly unlikely that two or more of these molecules induce the same promoter, we expected a high level of fluorescence with one molecule and low levels with other molecules. Some of the promoters followed this trend, while others showed a high level of fluorescence for the majority of molecules. The latter were disregarded since their fluorescence wasn’t in response to a particular molecule as the inducer.
Our team also noticed that some of the fluorescent plates had been inaccurately matched to their promoters. This could explain the lack of increasing fluorescence with increasing molecule concentration, the lower levels of fluorescence compared to the initial screening, and the lack of activation by a specific molecule. We corrected those promoters and performed the titration using the corrected strains.
The revised promoters improved the analysis of a promoter’s fluorescence level against all ten of the molecules. The revised promoters were largely only induced by one molecule. Based on our final data, we selected the molecule-promoter combinations that produced the best trend in fluorescence. These pairs have the potential to be further developed into biosensors for the molecules and greatly contribute to academic research. In particular, the following molecules and strains were identified to have an increasing trend with increasing concentration of the molecule: BHL with the ydel promoter produced a 1.7x fold increase, CND with the ybcK promoter produced a 2.2x fold increase, CND with the aegA promoter produced a 2.8x fold increase, and DEP with the yfiF promoter produced a 2.0x fold increase. We then further analyzed these strains through computational modeling and research.