Engineering

Introduction

The McMasterU team believes that adjustments are a fundamental part of research, providing the opportunity to learn from mistakes and refine ideas in order to generate novel products that benefit both the scientific community and the world at large. We put this into practice throughout the course of our project, where we designed a filamentous phage plasmid with the intention of producing phage particles to use in a vaccine to combat neglected parasitic diseases. Despite experiencing several obstacles while refining protocols and experiments in the process of constructing our plasmid, we were able to revise our errors and learn from them, granting us an even deeper understanding of our project. Through this learning curve, issues were troubleshooted and ultimately, breakthroughs were made with novel designs that helped our project reach the point it is at today.

Optimization of fdGPS2.1 (RGD-4C)-Amp plasmid

The centerpiece of our project was the design and construction of the fdGPS2.1 (RGD-4C)-Amp plasmid that would generate phage particles. This design had several steps and was not an easy task to accomplish, as designing it required several hours of commitment and hard work to generate a product that would potentially work in application. As such, our dry lab team worked extensively on the development of the fdGPS2.1 (RGD-4C)-Amp plasmid through a sequence of iterations (see Figures 1-6), crafting each segment of the plasmid with care to avoid potential errors that may arise during its construction.

Figure 1: Design schematic of fdGPS0.0.

Figure 2: Design schematic of fdGPS0.0 with ampicillin resistance genes.

Figure 3: Design schematic of fdGPS0.1(J23100) with ampicillin resistance genes.

Figure 4: Design schematic of fdGPS0.2(J23100, RGD-4C) with ampicillin resistance genes.

Figure 5: Design schematic of fdGPS1.2(J23100, RGD-4C) with ampicillin resistance genes.

Figure 6: Design schematic of fdGPS1.2(J23100, RGD-4C) with kanamycin resistance genes.

Naturally, errors arose during the design cycle, yielding a product that was not viable during experimentation. This was the case with the fdGPS1.2(J23100, RGD-4C)-Kan plasmid iteration, which was initially designed to ligate together using a Gibson assembly method. However, we found the use of six sequences made it challenging for the experiment to be successful, as it was observed that the plasmid failed to construct during the ligation component of the experiment. We changed our approach and modified the restriction enzyme used for cutting the sequences, as well as the cutting sites on each sequence, to allow us to run a Golden Gate assembly which can accommodate a greater number of sequences. The new design for our plasmid (fdGPS2.1 - Amp) had a total of six sequences, two of which were PCR generated, that would be ligated together during Golden Gate assembly.

Figure 7: Design schematic of fdGPS2.0 with ampicillin resistance genes. This schematic was generated with Golden Gate assembly overhang sequences instead of Gibson overhang sequences

Figure 8: Design schematic of fdGPS2.1 (RGD-4C) with ampicillin resistance genes (designed using Benchling). Plasmid designs developed by William Pihlainen-Bleecker

Optimization of Modeling

Modeling a biological response is a very big part of genetic engineering, especially for something with as far reaching effects as a vaccine that goes into our bodies. While we are a ways away from fully creating an efficient, safe, and effective vaccine, we have taken many steps to validate the design of our project. A more in depth explanation of our work and research is documented on other pages of our wiki, but we will provide a simplified overview here.

Immunological Response Model

Firstly, to select the right antigens to stimulate an immune response, we referred to literature to recommend what might be good options to combat Echinococcosis and Cysticercosis. We selected TSOL18 and EG95 as our antigens of choice, despite these antigens having not been widely used in vaccination in earlier studies. Since our AAVP relies on the selection of immunogenic epitope sequences from these antigens to display on the viral coat, a predictive simulation with The Immune Epitope Database (IEDB) was needed to identify the best sequences from both proteins. Simulations are an incredibly important part of engineering success, as they allow us to inform the construction of our project without having to actually test it in organisms, which can be very harmful and financially costly if unsuccessful.

Figure 9: Graph depicting the location of predicted immunogenic B cell epitopes (yellow regions) of the TSOL18 protein, obtained from an IEDB simulation (http://tools.iedb.org/bcell/).

We took advantage of additional tools by simulating an immune response to the antigens using online software. The results showed an inflammatory response, confirming immunogenicity of the selected antigens. However, this simulation performed better given the whole protein sequences instead of just the sequences of selected epitopes. We also used many default parameters for this simulation, as well as for the epitope prediction done earlier. Simulations are useful, but there is still much work to be done to get an accurate prediction such as specifying the amount of antigen to inject. We believe the most important thing about modeling is identifying these limitations and striving to create better models by building upon the work of others. By continuing our work on the AAVP and proceeding with further experiments in future engineering cycles, we will be able to validate the predictions of the models we use or suggest improvements.

Figure 10: A prediction of the expected antibody response upon injection of the TSOL18 and EG95 proteins. The black line depicts the level of antigen (the injected proteins) while the other lines indicate different types of antibodies. This graphic was obtained from a C-ImmSim simulation (Rapin et al. 2010).

Our approach to modeling the effect of our AAVP extended well beyond using software tools as the team conducted extensive research into the expected and desired immune response to the AAVP, the mode of entry of the AAVP into the body and specific cell types, as well as the optimal delivery method and materials required to get the AAVP to its target location. We consider this research part of our approach to modeling and to engineering success as an understanding of all these diverse aspects of our project is essential to informing the construction of our AAVP and predicting its efficacy. Our work on the AAVP is a reflection of the intertwined complexities existing in any field of biology.

Vaccine Distribution Model

It is important to be aware that projects in synthetic biology extend beyond just biology itself, as seen through our investigation of the optimal distribution strategy for our AAVP vaccine. For a vaccination campaign to be effective, it is essential to consider many parameters such as the quantity of vaccine available, who to distribute it to, and how quickly it can be distributed to name a few. This, in turn, is affected by determining who the most vulnerable demographics are, who is most likely to transmit the disease, how long the disease lasts, how long it takes for symptoms to appear, how severe the symptoms are, rates of transmission, and so on.

We come to see that vaccination depends on so much more than biological mechanisms – with socioeconomic and geopolitical factors immediately coming into play. Based on this, we searched for potential models to determine optimal vaccine allocation strategies which were often found to rely on data about vaccine efficacy, costs, and demographic information in order to produce meaningful output. Unfortunately, we are unable to reliably predict or determine this data at our project’s current stage, and thus attempted to recreate a simplified version of one such model (Wen et al., 2023). By deconstructing and rebuilding this model, we gained an understanding of the complexities of effective vaccination that extend beyond modeling synthetic biology. It provided us with an understanding of vaccine roll out based on the allocation of resources within a particular region, providing estimates of how many individuals would be given a dose within each priority group given the constraints of budget, storage capacity, transportation, and more as outlined in Figure 11. Throughout the course of this iGEM cycle, we gained an appreciation of the necessity of models for informing the optimal use of scientific innovations, particularly in areas that have been otherwise neglected. Moving forward, the collection of data such as the disease’s basic reproduction number, R0, in regions with an overlap of both diseases, such as Latin America, would add greater specificity to this optimization model for future global use.

Constraints used in the function to maximize output of vaccine distribution in a specific area, i, after identifying parameters such as priority groups, target doses, costs and budget, storage capacity, and the percentages of vaccinated and unvaccinated individuals in a population. Constraint (2) ensures that the number of vaccinated individuals should not exceed the number of target population in the area. Constraint (3) ensures that the vaccination ratio for each priority is not lower than the minimum pre-set coverage rate to prevent low vaccination rates. Constraint (4) indicates priority be given to those who have received less than the doses administered. Constraint (5) represents the relationship between the amount of vaccine allocated and the number of people vaccinated, considering vaccination history. Constraint (6) is the constraint that the total allocation of vaccines in all demand points does not exceed the stock in the allocation center and (7) is that the amount of vaccines obtained in each region does not exceed its warehouse capacity. Constraint (8) indicates that the total cost of vaccine distribution and vaccination in each region does not exceed the local budget.

Optimization of Experimental Protocols

Optimization of PCR protocol

Before carrying out our Golden Gate assembly, we first needed to optimize our PCR protocol to make sure that we had a sufficient concentration of PCR product for a successful plasmid construction. An analysis of our initial experiments revealed that the PCR yield of one of our plasmid fragments was lower than we needed. Through several PCR tests and gel electrophoresis analyses of the results we were able to optimize the protocol for the PCR of each individual fragment, particularly by systematically testing different temperatures and durations for the various steps of the PCR cycle. We discovered that the optimal temperature for functioning of each single stranded DNA primer ordered from IDT was approximately 57.5 degrees Celsius for the forward and reverse primers elongating our shorter fragment, and 70 degrees Celsius for the forward and reverse primers elongating our longer fragment. This was confirmed after running both primers at their respective temperatures and then observing the DNA samples on an agarose gel.

Optimization of Gel Electrophoresis protocol

After optimizing our PCR protocol, we turned our attention to improving how we ran our gel electrophoresis protocol, which generated smears while running (Figure 13). Upon further investigation, we found that the use of higher voltages was one of the issues that caused our bands to bend and smear as they traveled down the gel. To fix this, we reduced the voltage from 110 V to between 70-90 V, and increased the time the gel ran for to allow the bands to migrate far enough down the gel. Another resolution to the smearing was decreasing the volume of the loaded sample into each well. We discovered that we were adding too much sample to each well, which caused overflow and loss of some of our sample. Once we reduced our volumes to approximately 24 uL (8-well comb, 35 mL of agarose gel), the bands showed less smearing and distortions (Figure 14).

Figure 12: Gel electrophoresis using DNA samples. Smearing and distortions are present in the DNA ladder and samples.

Figure 13: Gel electrophoresis using DNA samples. Smearing and distortions have reduced significantly compared to the previous figure.

Optimization of Gel Purification protocol

Extraction of the DNA samples from the agarose gel presented a new problem for our team. Using a razor blade, we carefully removed a gel slice that contained our sample DNA, and proceeded to dissolve it using a Monarch Gel Extraction Kit (T1020S). After performing a nanoquant analysis, we realized the data we were obtaining was not viable for use in a Golden Gate assembly experiment. Despite trying several times with this purification method, our data did not yield results we could use for our plasmid construction. Therefore, we pivoted to using a Monarch DNA/PCR Cleanup Kit (T1030S) to purify our sample without having to extract it from agarose gel. Once we made this adjustment, we found that the data we observed (Table 1) were viable for a Golden Gate assembly, allowing us to push forward with our project.

Table 1: Nanoquant analysis results after DNA/PCR Cleanup of f3-55nm DNA fragments.

Optimization of Bacterial Transformation protocol

While practicing techniques for experiments, as well as performing experiments to generate data, we found that our bacterial transformation protocol needed to be modified. While performing the experiment, we found several instances where we were performing protocols either incorrectly or not as efficiently as we could have been. Such moments include heat shocking the bacteria cells for the appropriate amount of time, the time given for incubation after shock, and the concentration of antibiotics used for LB agar plates. Our original approach to performing transformations ultimately yielded results that were not viable, leading us to optimize the protocol to avoid future issues. These changes included heat shocking at 42 degrees Celsius for 30 seconds, incubating the bacteria cells for up to an hour, testing the concentration of ampicillin and tetracycline added to the LB agar plates and comparing them, adding a serial dilution step to test the optimal dilution factor, and implementing streaking to separate colonies (Figure 14). These changes proved to be useful when running future transformation experiments, allowing us to generate colonies that would be used for future research.

Figure 14: Streaked plates of NEB Stable Competent E. coli cells containing f3-55nm and fMCS plasmids.