Team:UChicago/Poster

University of Chicago GeneHackers: Optizyme 2020

Optizyme: A Software Package for Prototyping and Optimization of Cell Free Systems

Presented by the UChicago GeneHackers 2020

Patrick Sun¹, Avery Rosado¹, Michelle Awh¹, Michael Disare², Haneul Yoo², Will Grubbe², Jessica Priest², Kourtney Kroll², Frank Gao², Philip Ross², Matt Wash², Fabien Bylehn², Ben Glick³, Dmitri Kondrashov³

¹iGEM Student Team Member, ²Graduate Student Advisor, ³iGEM Team PI, Department of Molecular Biology, University of Chicago, Chicago, IL, USA

Abstract

Optizyme is an open source package in the computer language R that aims to streamline computational approaches to synthetic biology problems. Optizyme contains a suite of functions that allow for Michaelis-Menten model construction and visualization given only biological parameters. Most importantly, Optizyme allows for optimization of enzyme concentrations within cell-free systems through a gradient descent algorithm that accepts models built using Optizyme’s capabilities, as well as any user constructed model that fulfills certain requirements detailed in the package documentation. Accuracy of the optimization algorithm is tested on the model described in “A combined experimental and modelling approach for the Weimberg pathway optimisation” (Shen et al). Using their model, Optizyme identifies a higher performing optimum than the optimal ratio identified by Shen et al. We then apply the capabilities of Optizyme to optimize the design of a cell-free system we intend to construct next year that degrades polyethylene terephthalate into catechol.

Introduction

Computational approaches to biology are capable of leveraging the improving quality and quantity of experimental information consolidated in scientific databases. Furthermore, computational models and optimizers can explore a wider range of possibilities than can be experimentally explored in a lab. However, there does not yet exist a tool that streamlines the steps of the computational workflow. Our mission is to develop an accessible tool that facilitates computational modelling and optimization to accelerate the adoption of computational prototyping by the biology community.

Inspiration

The United Nations sustainable development goals (SDGs) lists 17 broad issues that plague humanity today, and are of the utmost importance to resolve. When brainstorming for our project, a quick glance through the UN SDGs inspired us to tailor our project towards addressing a few of these major issues. Specifically, our project aims to address the issue of microplastic pollution that poses a threat to marine life and biodiversity as well as human water and food sources. This year our team has used information about plastic degradation pathways, specifically polyethylene terephthalate (PET) degradation pathways, as well as other pathways found in biological metabolism, to construct a novel four enzyme pathway that converts PET into catechol, a biodegradable compound that is used in the synthesis of central nervous system drugs. We originally planned to construct this cell-free system in the lab for our project, but were prevented from doing so due to the COVID-19 pandemic. At this point, we pivoted our focus towards designing our system in the most efficient way possible, and that is when we began realizing how powerful computational approaches to this problem would be. Computers could consider significantly more possibilities through simulation than experimental approaches ever could. We devised a workflow that would work for not just us, but for any team of researchers that were engineering a cell-free system.

However, as we attempted to apply this workflow to model and optimize our PET degradation pathway, we quickly realized that there was no software that streamlined this process for us. If the biology community was to be able to leverage all the benefits of computational techniques, then there would have to be a software tool that made this workflow more accessible to researchers, and we set out to create this tool.

Project Goals

Our software aims to make computational prototyping more accessible to scientists in industry. We hope that the increased ease-of-use in applying computational methods that Optizyme offers will allow researchers to more efficiently optimize industrial applications of cell-free systems. Because Optizyme is built for general use in designing cell free systems, it has potential applications that span diverse fields from biomanufacturing to pharmaceutical synthesis, and could dramatically facilitate efforts in all of these fields.

Design

Optizyme uses a classic biochemical approach to model the activity of cell-free systems known as Michaelis-Menten kinetics, which is a well known differential equation framework for modelling enzyme kinetics whose necessary parameters are often reported in the literature. However, Optizyme focuses only on enzyme concentrations and final product yield predicted by a Michaelis-Menten simulation to distill it down into a class of functions known as multivariate scalar functions.

By restructuring the information in a Michaelis-Menten simulation into this form, Optizyme is able to use Michaelis-Menten simulations to explore and optimize the activity of a cell-free system using a gradient descent algorithm. The algorithm uses theorems of calculus to run Michaelis-Menten simulations and intelligently generate a mathematical “compass” that it uses to navigate through possible enzyme concentrations as it works to optimize the activity of the cell free system.

P is the current best solution found by the algorithm, while P in the next step will be an improved solution. Gamma represents differential step size and nabla F represents gradient vector of the function generated by the Michaelis-Menten simulations.

Engineering

Optizyme 0.1.0: Contained one optimization function, no modelling capabilities, and required the user to construct a modelling function that fit certain requirements that made it readable by Optizyme.

Optizyme 1.0.0: The optimization capabilities of the software is improved to consider physical constraints present in a wet lab environment.

Optizyme 1.1.0: Model construction and visualization functions added to Optizyme. Modelling function is compatible with Optizyme’s algorithm for fully self-contained computational capabilities within Optizyme.

The versions of Optizyme have evolved to fit our needs for the software, as well as through incorporation of feedback from experts in industry and our envisioned end users, which will be discussed in more detail under “Integrated Human Practices”.

Our original plan this year was to use Optizyme to computationally model and optimize our PET degradation pathway. Next year, we would then clone the necessary genes into E. Coli and then confirm that we had successfully incorporated the genes into E. coli. We would then lyse the cells and collect the resulting lysate to construct our cell free system, and we would be able to combine enzymes in the concentrations determined by Optizyme to construct our system to be as efficient as possible.

The QR codes included are the QR codes to Optizyme's Github page.

Proof of Concept

We demonstrate a proof of concept for our optimization algorithm through working with the paper “A Combined Experimental and Modelling Approach for the Weimberg Pathway Optimization” (Shen et al. 2020). Within this paper, the authors worked to optimize the Weimberg pathway, which is a five enzyme pathway that converts D-xylose to α-ketoglutarate, and is depicted below (Shen et al. 2020).

The authors performed initial-rate experiments to determine the kinetic constants and reaction mechanisms of each of the enzymes involved in the pathways, and then used these findings to construct a model based on the system of differential equations that represent the time evolution of their cell-free system. The researchers constructed their cell-free system in the lab, and compared the time evolution of the real system with the time evolution predicted by the model to confirm that their model accurately predicted the effect of the system. The authors then used their model to computationally optimize the ratio of enzyme concentrations while holding the total enzyme concentration fixed at 22.6 micrograms/milliliter. The authors’ reported optimal ratio of enzymes results in a time to reach 99% yield of 63.07 minutes where the optimal enzyme concentrations were 2: 1.5: 5.7: 9.8: 3.6, where the concentrations are in micrograms/milliliter. We reconstructed the model presented in the paper, and used the optimization algorithm included in Optizyme to determine the optimal ratio of enzyme concentrations, which we determined to be 2.1: 1.6: 5.9: 9.3: 3.7. Our optimal ratio of enzyme concentrations gives a time to reach 99% yield of 62.85 minutes, a slight improvement from the answer found in the paper. The answer achieved by Optizyme is qualitatively the same as the solution found in the paper, but faster time to reach 99% yield in our solution demonstrates that Optizyme’s algorithm is capable of identifying improved optima compared to existing methods. In the graph on the left, the activity of the system predicted to be optimal by Optizyme is compared to that of the optimal identified by Shen et al. We see that qualitatively the systems behave the same, and this is supported by the barplot on the right, which shows that the final yields for the optimal systems identified by both Optizyme and Shen et al are nearly identical.

Integrated Human Practices

Michael Köpke PhD:

Feedback: Model construction may be a limiting step in computational optimization.

Integration: Michaelis-Menten modelling capabilities for enzyme pathways of arbitrary length that account for multi-substrate enzymes, competitive inhibition, and noncompetitive inhibition are integrated into Optizyme.

Michael Jewett PhD:

Feedback: Depending on the units used, experimenters may only be able to control enzyme concentrations to 1 or 2 significant digits. Also, our software should be accessible to as many people as possible.

Integration: Flexible user control of algorithm accuracy is introduced to Optizyme. The algorithm efficiency is improved by removing unnecessary computational steps associated with optimization of excessively low concentrations. Optizyme is designed to require no mathematical inputs.

Science Communications

We are designed and are facilitating three-part workshops for two New York City education initiatives: You Can Too and REACH, run by Columbia University and Regis High school, respectively. This curriculum spans Autumn Quarter and provides students the opportunity to apply synthetic biology to solve real world problems.

Our Curriculum Entails:

Making SynBio: An introduction to synthetic biology, its capabilities, and applications.

Environmental Issues: Exploration of environmental issues and the current initiatives to combat them.

Creating a Solution: Students synthesize what they’ve learned throughout the quarter to design a bioremediation solution.

This is a student-focused approach that allows kids to take a leading role in the solution design process, which we hope will inspire the next generation of scientists and iGEMers to tackle the world’s problems through SynBio.

Modelling and Optimization

The greatest sites of plastic pollution are the garbage patches in the world’s oceans, so the most efficient places for our cell-free system to be employed is in these pollution hotspots.To begin modelling the efficiency of our cell-free system we first did research on what amounts of microplastic the system would be employed to deal with. Isobe et al reported on the density of microplastic particles per unit volume in the ocean and the data point with the highest microplastic content was approximately one piece (<5mm diameter) per cubic meter. Approximating the particles as perfect spheres allows us to estimate the ocean microplastic density as .723 grams of microplastic per cubic meter. This is equivalent to roughly .003763 moles of PET per cubic meter, or .000003763 moles of PET per liter. This is the first scenario that we will model: the use of our system to remediate naturally occurring densities of microplastic. In all the models that we will construct, we hold enzyme concentrations constant at 1 micromolar, which is a concentration that is certainly achievable in the lab.

Our algorithm maximizes the product yield after a given amount of time. The results of our optimization for each scenario is compared to a non optimized system with a 1:1 ratio of all enzymes, as well as a system that is constructed with enzyme ratios proportional to inverse turnover rates. In each of the line graphs representing activity of our system, the green line represents the activity of the system Optimized using Optizyme, the red line represents an arbitrary 1:1 ratio, and the blue line represent enzyme ratios proportional to inverse turnover rate. For each time-course, a bar graph is included below that illustrates the yield of final product for each of the systems.

System activity is also explored for PET concentrations spanning three orders of magnitude: .001M, .01M, and .1M. These concentrations are modelled and optimized to explore the activity of our cell free system if it is implemented in a setting with higher than naturally occuring PET concentrations. Similar to above, time courses are represented and their corresponding final yields are depicted in bar plots below them.

References and Acknowledgements

The UChicago GeneHackers is able to fulfill its mission of pioneering innovative solutions to pressing matters through research in synthetic biology thanks to the hard work and commitment of our full team, from undergraduate researchers to graduate advisors to our extremely supportive principal investigators.

Avery Rosado: Summer team member; Outreach, Science Communication, Collaboration, Wiki, design

Michelle Awh: Summer team member; Turning the Optizyme algorithm into a package on Github, Science Communication

Patrick Sun: Summer team member; Algorithm design and implementation, Collaboration

Jessica Oros: Academic year member; project idea, outreach

Angela Marroquin: Academic year member; project idea

Nathan Sattah: Co-president of GeneHackers; planning, summer-team organization and oversight

Sneha Kesaraju: Co-president of GeneHackers; planning, summer-team organization and oversight

Rachael Filzen: Past-president of GeneHackers; planning, summer-team organization and oversight, oversight of transition into new administration

Special Thanks:

The work completed by GeneHackers over the course of the academic year and the summer is the result of contributions made by a host of devoted individuals from various areas within the University of Chicago. All work carried out to successfully design and carry out this year’s dry-lab work was completed under the supervision of primary investigators Professor Ben Glick and Professor Dmitry Kondrashov who offered their feedback and advice. Additionally, graduate student advisors Haneul Yoo, Kourtney Kroll, Jessica Priest, Michael Disare, Philipp Ross, William Grubbe, and Frank Gao met with the summer team and the GeneHackers board members on a weekly basis throughout the summer to offer their insight and advice into refining workflow and steering the project forward. These meetings were facilitated by the year-round GeneHackers board with oversight from Co-President Rachael Filzen, who handed leadership over to newly elected Co-Presidents Nathan Sattah and Sneha Kesaraju during the summer months.

Despite the team’s transition to a wet-lab environment in response to COVID-19 related lab closures and limited access to university facilities, all work was completed remotely in a dry-lab environment by GeneHackers members, who carried out coding, field-work, and Wiki-related work and coordinated using thorough virtual means.

As the team worked to overcome the challenges presented by the pandemic, learning from experts in real-world, applicable fields became a vital component of our work. Professor Michael Jewett of Northwestern University, who played a crucial role in the design of the iPROBE system for optimizing biosynthetic pathways via cell-free media, provided his insight for making the structure of an early iteration of the Optizyme software tool more intuitive and efficient. A later meeting with Dr. Michael Koepke, Vice President of Synthetic Biology at LanzaTech, and members of his team tailor our algorithm to our envisioned end user experience thanks to their suggestions for boosting modelling capabilities.

Boston University PhD candidate Chris Kuffner offered input based on background in PET degradation and advised the team in navigating the logistics of the iGEM competition.

Generous stipend funding was provided by the Biological Sciences Collegiate Division of the University of Chicago and the Pritzker School of Molecular Engineering at the University of Chicago.

Sources Cited:

Shen, L.; Kohlhaas, M.; Enoki, J.; Meier, R.; Schönenberger, B.; Wohlgemuth, R.; Kourist, R.; Niemeyer, F.; Niekerk, D. van; Bräsen, C.; Niemeyer, J.; Snoep, J.; Siebers, B. A combined experimental and modelling approach for the Weimberg pathway optimisation. https://www.nature.com/articles/s41467-020-14830-y (accessed Oct 27, 2020).

(2020). Retrieved November 07, 2020, from http://jewettlab.northwestern.edu/team-members/michael-c-jewett/

(2020). Retrieved November 07, 2020, from https://www.lanzatech.com/

Isobe, A.; Iwasaki, S.; Uchida, K.; Tokai, T. Abundance of non-conservative microplastics in the upper ocean from 1957 to 2066. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6345988/ (accessed Oct 28, 2020).