Team:Calgary/Model
Overview
We use models to theoretically describe systems. This description can be used for prediction, monitoring and control, and further optimization -- all of which are essential in translating our proposed solution from theory to praxis. In addition to advancing the field, model development and implementation are used as resource-effective tools, which are especially important in the iGEM landscape. Furthermore, constraints engendered by the pandemic only highlighted the need for modelling when access to resources are limited.
Modelling Goals
Throughout the project, the following areas of inquiry arose:
How well does an adsorber column packed with lanmodulin-functionalized cellulose beads perform? How can we modify the design and process variables to optimize performance?
What effect does the length or type of a linker have on a fusion protein? How can we improve our measurement constructs? How can we create a workflow for optimizing linkers when developing fusion proteins? How can we better characterize the fusion proteins we plan on producing?
How to improve the lanthanide binding capacity of lanmodulin through increased binding affinity?
How to activate lanmodulin’s pseudo binding pocket to allow for four active binding pockets?
How can we provide better estimated values to our wetlab?
To address these questions, we employed the following models.
Packed Adsorber Column for Lanthanides (PACLan) Performance Model
Despite lanmodulin’s promise in prior literature, we did not know how well it would perform when it is immobilized on cellulose beads and packed into an adsorber column. Thus, we modeled the performance of this high-throughput system using literature and predicted values. We also realized that these preliminary values may not lend itself to optimal performance, so we used a parametric sweep to determine the values that will optimize the system. Learn more here.
Optimal Linkers
Although inspired by literature, we were quick to realize that our team needed to produce the most optimal measurement constructs to reduce the amount of trial and error required to successfully produce a rare earth element detection system. Using the engineering cycle, this project documents the revisions and improvements we made to choose the optimal linker for each novel protein. Additionally, this project is a contribution to those with the desire to create their own novel proteins. Learn more here.
Mutant Protein Library
While lanmodulin presents a very impressive lanthanide binding capacity, in order to compensate for the potential of reduced affinity within cellulose bead-lanmodulin fusion complexes, our team developed a mutant protein library of lanmodulin. The mutant library was then assessed for binding affinity to find sequences with the greatest increase. Learn more here.
Genetic Algorithm
One of lanmodulin’s areas with the greatest potential for improvement lies in its fourth pseudo binding pocket. Unlike its three other binding pockets, this one is not able to actively bind lanthanides, which opens up the potential for a mutated lanmodulin which binds up to 4 lanthanides. Our team used a 3 layered EF-Hand mutation workflow to develop an optimized fourth binding pocket using a genetic algorithm with an intelligently initialized set of EF-hand sequences made through an amino acid family conservation program and a probability averaging model using a database of other EF-hands. Learn more here.
Protein Subspace Exploration
Every amino acid sequence has a subspace of all potential mutations, however making an exhaustive list of all these permutations for testing mutations is not only computationally expensive, but it becomes entirely unfeasible the longer protein sequences get. Our team designed 4 different mutation systems to modify lanmodulin by accessing different sample spaces of the total subspace. Learn more here.
Pessimistic Optimisitic
The Ideals of Modelling Ideally
Computational models are very useful tools, but they often fall short of being representative of the real world. To make our models more representative of real world conditions, we integrated a pessimistic-optimistic workflow into our system designs. In effect, by anticipating a worst case and a best case scenario for a given model, we can better anticipate its performance in the non ideal range of conditions our proteins could be exposed to. Basically, we want our systems to struggle while they are still models so they can do well in the lab.
Fusion proteins, the good, the bad, and the pessimistic
An example of this workflow in action is seen in our luciferase fusion protein modelling (link to the measurement modelling page). To give a quick summary, we wanted to see what length of spacers between lanmodulin and our light emitting proteins would produce the best overall construct, where the luciferase proteins had the smallest amount of space between them, and lanmodulin was the least denatured. To do this we made homology models of the fusion proteins with Chimera, then chose our optimal and least optimal models by comparing different homology model criteria values. By visual inspection, we chose the models with the least and most amount of space between the active sites of the luciferase proteins, which were attached to either end of lanmodulin. These two options would act as our pessimistic and optimistic choices for each set of spacers, allowing us to average the two and get a more reliable value for the spacing we could expect in our fusion protein. Using those averaged values, we filtered out the majority of the spacer combinations and were left with a handful of potential fusion protein candidates. Finally, the winning model was chosen by visual inspection, as lanmodulin’s tertiary structure was the least denatured in that model. Bind pocket optimization is about increasing the average number of lanthanides binding, not just increasing binding affinity.
Expecting the Unexpected
Lanmodulin is a very recently discovered protein, and as a result there are many unknowns about it’s behaviour. For instance, the protein’s binding pockets might experience reduced binding affinity when immobilized due to some kind of denaturing, which can be problematic for our proposed rare earth element recovery system. To get ahead of this, our team also developed multiple potential lanmodulin mutants which improve on binding capacity in different ways. Our two main approaches involve functionalizing a pseudo binding pocket in lanmodulin, and optimizing overall protein binding affinity through the mutation of structural hotspots. Both of these approaches aim to increase the average number of lanthanides lanmodulin is capable of binding, which should enable us to overcome the potential negative impacts of immobilization.
