Project Abstract and Inspiration

Project Description

Genetic Design

Back to Basics: Standard Parts

The following parts are the basic components of our projects including the promoter, terminator, ribosome binding site and markers for our project. These parts are essentially the framework of our project.

1. pSB1C3: A plasmid backbone with chloramphenicol resistance
2. BBa_J23100: Constitutive Promoter
3. BBa_J23102: Second Constitutive Promoter
4. BBa_B0030: Ribosome Binding Site
5. BBa_B0010: Terminator
6. BBa_E0040: Green Fluorescent Protein (GFP)

Ecstatic Enzymes and Marvelous Mutations: Converting Caffeine to 7-methylxanthine and Paraxanthine

To test the efficiencies of NdmA and NdmB in-house and focus on raising the efficiency of NdmB via mutations, wet lab created new basic parts of the genes of focus found in BBa_K1627006 and attached his-tags for our affinity chromatography step. Below are the parts used in our project with “α” signifying parts in our original plasmid and “β” signifying parts in our improved plasmid. By adding the sequence of BBa_K458003 into the original operon from UT Austin in BBa_K1627006, the improved catalytic efficiency of NdmB is brought into their part as well.

1. BBa_K4580000^αβ: NdmA-6x his
2. BBa_K4580001^α: NdmB-6x his
3. BBa_K4580002^αβ: NdmD-6x his
4. BBa_K4580003^β: αNdmB-6x his (best mutated version)
5. BBa_K4580004: Composite part of NdmB-6x his and GFP

Below you will find the original plasmid we designed and the updated, optimized pENGM42 plasmid constructed later on.

Wet Lab

Introduction

In 2012, the University of Texas at Austin (UT Austin) utilized the custom plasmid pDCAF3 to generate the operon NdmABCD (BBA_K734000) and accurately measure the amount of caffeine in different solutions by taking advantage of the ΔguaB E. Coli strain. Team Cornell will be repurposing the genes found in pDCAF3 for our 2023 project ENERGEM in our pENGM42 plasmid. This project relies on applied biological/biomedical engineering and in silico modeling as we build a strong product foundation. Thus, there are three major wet lab focuses:

1. Adaptive/Directed Evolution (ADE)
2. Enzyme Immobilization via Encapsulation
3. in silico modeling of epPCR and fluid flow within the IBCS reactor

Pre-Adaptive / Directed Evolution: epPCR

It is important to maximize the catalytic efficiency of the enzymes of interest since ENERGEM revolves around efficient production of methylxanthines. The three enzymes included from pDCAF3 are NdmA, NdmB, and NdmD. These enzymes are part of the bacterial caffeine metabolism III pathway, with functions of each enzyme illustrated below:

NdmA has a catalytic efficiency (K_cat/K_m) of 5.1 ± 1.2 min^-1 µM^-1 while NdmB has a catalytic efficiency of 0.006 ± 0.001 min^-1 µM^-1. Thus, increasing the catalytic efficiency for NdmB is important for our goal of efficient methylxanthine production. There is no documented catalytic efficiency of NdmD as its purpose is to function as a coenzyme (transfers H⁺ from NADP(H) to catalyze demethylation).

The first step in our method of choice to increase K_cat for NdmB is inducing mutations using error prone PCR (epPCR). This method involves changing the composition of PCR reactions which decreases the fidelity of DNA polymerases and increases the rate of mutation, a fact we utilized to alter NdmB. While SNPs (single nucleotide polymorphisms) tend to be random, epPCR’s specificity can be increased by incorporating two strategies. The first is to design primers to target a specific region of DNA. The second is to use an experimentally determined amount of manganese chloride (MnCl₂) to alter the PCR reaction composition and decrease the fidelity of the DNA polymerase (Q5® High Fidelity by NEBioLabs). The prevalence of generated mutations differs based on the concentration of MnCl₂ in a linear fashion (i.e. more MnCl₂ = more mutations). The following table illustrates the rate of mutation per DNA template based on its length and amount of doublings [Link].

# of Doublings	Length of DNA (bp)					Mutations per Nucleotide Position
	100	200	400	800	1600
5	0.33	0.66	1.3	2.6	5.3	0.0033
10	0.66	1.3	2.6	5.3	11	0.0066
20	1.3	2.6	5.3	11	21	0.013
30	2.0	4.0	7.9	16	32	0.02
50	3.3	6.6	13	26	53	0.033

All data in the above table is obtained when [MnCl₂] is 25 mM. Importantly, the rate of mutation per 100 bp will increase linearly with the increase in [MnCl₂]. By targeting the region of epPCR as well as the rate of mutation, we can significantly increase the specificity of our modified epPCR cycle. This technique will generate a large library of mutants which we will clone into BL21 E. Coli for our adaptive/directed evolution cycle and imitate evolution to view which mutations accomplish an improved efficiency.

Adaptive/Directed Evolution (ADE)

Adaptive and directed evolution are methods to imitate evolution in vitro under a selective pressure. This pressure warrants the need of a plasmid that will help combat that pressure. To start our epPCR cycle, we ran epPCR on the entire plasmid, incorporating the two aforementioned strategies for specific epPCR. We used Takashi Koyanagi et al.’s method of ligation-free epPCR + plasmid assembly [4]. An important consideration of epPCR is that all mutations are randomly generated; there will be slight differences between two copies of the same template strand. To this end, we used adaptive evolution (AE), a method in which mutation will be induced only once. After cloning the plasmid into BL21 E. Coli, it will be grown in liquid cultures containing caffeine/paraxanthine. Adding these two compounds serves two purposes:

1. Bacteria will metabolize these compounds as our plasmid pENGM42 will be present
2. Caffeine is toxic to E. Coli and serves as a selective agent. Cultures with more growth (especially under paraxanthine treated conditions) will have a more efficient enzyme

The following diagram illustrates Team Cornell’s method of adaptive evolution via epPCR:

The purpose of adaptive evolution’s miniprep step will be to elucidate whether the DNA of bacterial colonies are becoming homogeneous. While generating many mutants is helpful, this is not a fruitful method when establishing data regarding Kcat, Km, and catalytic efficiency. With a homogeneous pool of DNA in a specific well, we are confident that the DNA sequence of the new and improved NdmB is representative of an entire well. Thus, miniprepping the best growers (step 4) is relevant for sequencing mutants.

However, the power of epPCR lies within its ability to generate large libraries of mutants. To this end, we implemented a cycle of directed evolution (DE) that runs concurrently with adaptive evolution. After an initial run of AE, we began a cycle of DE, as illustrated by the following diagram:

There are two important changes in the DE cycle. Firstly, epPCR occurs at the end of every growth period. This step continually induces new mutations in both the best growers and a new pool of template DNA. While the AE cycle ensures homogeneity, the DE cycle creates new mutants that could be better than previous iterations. It is also important to note that AE is inherently part of DE, the major difference being that epPCR occurs at the end of every DE cycle and only once at the beginning of an AE cycle.

After the completion of ADE, all collected samples were sent for DNA sequencing (Cornell University, Sanger Sequencing, NGS by Illumina, and PlasmidSaurus. The best grower contained the most efficient version of NdmB (denoted as αNdmB).

In-silico Modeling of epPCR and Fluid Flow in Reactor

To streamline the design process of our IBCS reactor, ANSYS Fluent and KinSim were used to analyze the thermodynamics, fluid flow, and reaction mechanics. As described further on the modeling page, analyzing these characteristics of the IBCS reactor will minimize prototyping, maximize reaction efficiency, and create a blueprint for industrial scale versions of this reactor. While fluid flow (both steady-state and turbulent) and thermodynamics. were analyzed by our product development subteam, the wet lab focus was on reaction mechanics.

The conversion of caffeine to methylxanthines is an endothermic process. Furthermore, the reaction is most efficient at 30℃. While a Sous Vide machine will be essential to maintain isothermal conditions, mitigating the amount of heat loss will ensure the reaction remains efficient over long periods of time. The energy equation can be modeled on ANSYS Fluent. We repurposed this equation to model heat loss and, as explained on the modeling page, specified the alginate beads containing enzymes as heat sinks. Therefore, fluid flow will be simultaneously simulated, generating a comprehensive model of the IBCS reactor. This model will be useful as a focus of the product development subteam is to generate a prototype for industrial scale-up. The method in which isothermal conditions are maintained for large-scale IBCS reactors will be influenced by the effectiveness of a Sous Vide machine in small scale prototypes.

Error-prone PCR is a method to introduce mutations into DNA sequences. However, due to the random nature of mutations that are catalyzed by epPCR, it becomes important to understand how mutations would affect the DNA sequence. WE developed 5 models in RStudio: defining a basic epPCR function in RStudio based on mutation rate, modifying the function predict [MnCl₂] using linear interpolation and nonlinear regression, and increaing the complexity by introducing different mutation types, mainly substitution, insertion, and deletion.

Our focus was in understanding the intricate balance between mutation generation and the likelihood of non-functional variants. By simulating epPCR in silico, we hope to mitigate the challenges associated with random mutations, paving the way for directed evolution experiments. The basis of our model lies in data from a study performed by David Wilson and Tony Keefe in 2000 discussing “Random Mutagenesis by PCR” [Link]. They found that the prevalence of mutations differs based on [MnCl₂], number of doublings, and length of DNA in a linear fashion [Link]. All models, except for the third one, are subject to the linear assumption. This assumption states that there is a linear relationship between the number of doublings and the rate of mutation per nucleotide position.

Model 1 develops the function error_prone_pcr which focuses on the effect of substitution mutations on DNA sequences. To validate our approach, we tested epPCR simulations with 10, 20, and 50 doublings, resulting in higher mutation rates and increased genetic diversity. We ran dummy sequences and the full sequence of our plasmid pENGM42 through error_prone_pcr with a mutation rate of 0.0033 100 times. This figure illustrates the number of mutations after 5 doublings:

Model 2 developed predict_mncl2_concentrations, a function that generates a predicted concentration of MnCl₂ to use for epPCR trials based on linear interpolation. Though biological systems tend to follow nonlinear behaviour, the data published by Wilson and Keefe validates our decision to run this simulation under the linear assumption. Below you can see the linear relationship between predicted [MnCl₂] and mutation rate per nucleotide position generated by this model:

Model 3 developed a nonlinear regression for predict_mncl2_concentrations. This model is designed to more accurately predict [MnCl₂] given a defined coefficient for doublings and coefficient for desired mutation rates. Interestingly, a linear relationship was established between mutation rate per nucleotide position and [MnCl₂], though the values of [MnCl₂] are different between model 2 and 3.

Model 4 developed simulate_mutations, a function that not only mutates DNA seqeuences via substitutions, but also considers insertions and deletions (indels). Indels typically cause frameshit mutations which are important to consider since a single frameshift mutation can cause that mutant to become non-functional. We found that if we run epPCR on our full plasmid, we can generate an optimal number of mutations without introducing frameshift mutations if the rate of mutation per nucleotide position is 0.00033 and number of doublings = 30. A one-way ANOVA confirmed this result.

Model 5 develops the basis of an upgraded version of simulate_mutations. This new, sophisticated version considers the effect of altered DNA sequences on the amino acid sequence of a mutant. Though model 4 indicated that the number of frameshifts is very close to 0, it is difficult to tell if the amino acid sequence was severely affected. Model 5 produces the following results within a modular function:

1. Original vs Mutated DNA sequence
2. Original vs Mutated amino acid sequence
3. Rate of substitutions, insertions, deletions, accumulated frameshift mutationsk
4. Functionality of mutant based on percent difference between original and mutant
5. Estimated [MnCl₂]

Hardware

The hardware involved in the synthesis of seven-methylxanthines from caffeine molecules is essential to the function and implementation of ENERGEM in the real world. The process of creating the chemical reaction involved in ENERGEM requires the use of free-flowing, individually created enzymes in a way that maximizes both efficiency and simplicity.

This efficiency and simplicity is achieved by immobilizing the originally free-flowing enzymes in two distinct systems, so that they can be easily reused, replaced, moved between labs, and tested. The first immobilization process involves encapsulating the enzymes in sodium alginate beads. These beads are then immobilized in a bioreactor, where the caffeine molecules flow freely through.

Enzyme Encapsulation

Methylxanthines are relevant to the pharmaceutical industry due to their utility in conditions related to myopia. Sterility of the final product, as well as any equipment used to create the product, is of utmost importance. For this reason, we will not use bacteria in the product development subteam’s isothermal biochemical synthesis (IBCS) reactor, explained in the Hardware section below. NdmA and NdmD, and NdmB and NdmD will be encapsulated in sodium alginate beads. The extrusion process of encapsulation is picture below:

Encapsulation allows the enzyme to be reused, regenerated, and replenished as the IBCS reactor produces methylxanthines at larger scales. These beads naturally take a spherical shape, increasing the surface area of the IBCS reactor.

The reaction occurs according to the following double replacement reaction:

As shown above, sodium alginate beads form when sodium alginate and calcium ions cross-link in solution. A shell forms and thickens quickly in solution. A small amount of the solution will be trapped within the bead after cross-linking. In ENERGEM, that liquid solution is the isolated NdmA/NdmD and NdmB/NdmD enzymes from the pENGM42 plasmid and therefore encapsulated by including concentrations of them in calcium chloride solution.

After a short incubation period, the encapsulated enzymes were analyzed to collect relevant data. The following equations were used to collect information regarding mean bead volume, diameter, and immobilization efficiency:

Mean Bead Volume:

Diameter of Beads:

Immobilization Efficiency

The beads were also observed qualitatively for uniform size, shape, consistency, and concentration.

Through a series of experiments, it was observed that the concentration of the calcium chloride solution into which the sodium alginate was dropped had no significant effect on any bead properties. For simplicity’s sake, we chose to utilize a 1 M concentration for all future batches.

Further experimentation found that sodium alginate concentration influences the viscosity of the sodium alginate itself, as well as the physical properties of the beads, including ability to deform without splitting under pressure and bead shape. We determined that our beads should be formed with a 2.5% weight by volume sodium alginate concentration, as they were sufficiently durable yet the alginate itself was still fluid enough to work with easily.

Bead Immobilization

The beads described above are mechanically immobilized within a 3D-printed framework of our own design. This design was developed through a series of drafts, and incorporates what we learned about the sodium alginate beads as we made more of them into our goals for the reactor.

Many drafts of the bead immobilization system were made, including designs that restricted movement of the beads horizontally and vertically, as well as designs that assumed the beads would come to rest in the framework and would only need to be restricted horizontally.

The design that was decided upon to immobilize the enzyme beads is a helix. The spaces between each spiral fit just one bead, allowing us to create spiral stacks of beads throughout the entire reactor.

This layout restricts beads both horizontally and vertically. Vertical restriction is required because, through testing, we discovered that beads would not always remain at the bottom of our bioreactor while reactant flowed through it. Total immobilization of these beads is superior to allowing the beads to float freely in the reactor because it allows us to assume residence time is consistent for all molecules, and thus reaction time within each bead is the same. Furthermore, immobilization of the beads will optimize flow and residence time while we pursue scale-up.

Bioreactor

Our goal with the design of the bioreactor is to create a fixed-bed bioreactor that can be easily scaled up to meet demand for the end product. The way we approached this is to use an easily-acquired housing combined with a specialized, but standardized, immobilization method. In the case of the former, we opted to use a 10mL syringe tube as our housing. The latter is satisfied by the immobilization framework we designed. The emphasis on scaling allows us to add an element of modularity and flexibility to the process, which in turn allows us to create an overall more efficient process for the creation of target products.

Human Practices

Our goal for this season was to determine the best possible field that our product could be used in, ensuring that we were meeting a pertinent need in the community, all while sharing our knowledge on synthetic biology and methylxanthines with our local Ithaca community and beyond. We started out the season doing research into the areas of methylxanthine use and found that historically, they have been used for respiratory illnesses such as asthma and COPD. However, as the season went on and we consulted more stakeholders in these fields, we learned that while 7-methylxanthines have been used for respiratory illness treatment, they no longer are used to the same extent as they once were. We used this feedback to pivot the focus of our project.

Our team began exploring other avenues to apply our work when we learned that there is ongoing research on how 7-methylxanthines can be used for myopia, specifically childhood myopia treatments. After our interview with Dr. Jingtai Cao, we learned that researchers and physicians have been looking into biological treatments for myopia as traditional surgical interventions are ineffective before the age of 21, supporting that our team was heading down the right path.

Through our interviews with a wide range of professors, therapeutics company leaders, physicians, pharmacists, and more, we learned that methylxanthines, as a whole, are extremely important compounds with a wide range of applications ranging from the medical industry to the agricultural field. Using similar mechanisms as us, we hope that others can build upon our work to produce different derivatives of caffeine. To aid in this process, our team developed a handbook with information on the wide variety of applications for methylxanthines. We expanded our education beyond just those who would actually use our product, to children in our community, by hosting events like the Sciencenter where we taught children about spherification through Orbeez, or events like Splash where we held bioethics debates on synthetic biology for high school students. Our team also hosted a summer science experiment series where we posted videos of ourselves doing experiments related to important biology concepts for kids. For instance, for ENERGEM, we decided to make popping boba or sodium alginate beads, as a kid-friendly version of our enzyme encapsulation system. We also reached out to the older members of our community, going to nursing homes like Kendal @ Ithaca and Longview Senior Living Community, presenting case studies including one of ENERGEM, gaining their insight on what they think the role of synthetic biology should be within the greater community. We hope that every event was as much of a learning experience for our attendants as it was for us.

In all, Cornell iGEM hopes that by providing a cost-effective, efficient method of producing 7-methylxanthines we can help provide our community with alternative and accessible forms for treatment for myopia.

References

[1] Summers, R. M., Louie, T. M., Yu, C. L., Gakhar, L., Louie, K. C., & Subramanian, M. (2012). Novel, highly specific N-demethylases enable bacteria to live on caffeine and related purine alkaloids. Journal of bacteriology, 194(8), 2041–2049. https://doi.org/10.1128/JB.06637-11

[2] Mock, M. B., & Summers, R. M. (2023). Mixed culture biocatalytic production of the high-value biochemical 7-methylxanthine. Journal of Biological Engineering, 17(1). https://doi.org/10.1186/s13036-022-00316-6

[3] Takashi KOYANAGI, Erina YOSHIDA, Hiromichi MINAMI, Takane KATAYAMA & Hidehiko KUMAGAI (2008) A Rapid, Simple, and Effective Method of Constructing a Randomly Mutagenized Plasmid Library Free from Ligation, Bioscience, Biotechnology, and Biochemistry, 72:4, 1134-1137, DOI: 10.1271/bbb.70814