Introduction

As a team, we aim to develop a project that not only works effectively but also has the potential to evolve into a successful business. With this in mind, our model idea was born. Using COBRApy (COnstraints-Based Reconstruction and Analysis for Python), we created a model to help scale up our holdfast production [1]. With our bioreactor in mind and our aspiration of a future business, this model lays the groundwork for building a bigger-scale system.

Motivation

Scaling up the production of microbiological products, particularly in a start-up environment, presents significant challenges. One of the most pressing concerns is reducing production costs while ensuring the efficient use of substrate materials, which are crucial for sustainability and long-term success. Optimizing the growth medium is a key strategy to achieve these objectives.

By fine-tuning the composition of the growth medium, we can dramatically enhance microbial growth and productivity, leading to a more cost-effective and resource-efficient process. In an environment where every dollar counts and time is often limited, the ability to maximize output while minimizing waste becomes not just an advantage but a necessity. Optimizing the medium allows for a balance between biological efficiency and economic feasibility, laying the groundwork for scalable success. Instead of performing an array of costly and time-consuming experiments, we utilized Flux Balance Analysis (FBA) on a genome-scale metabolic model (GEM) to predict the optimal conditions of our reactions.

Model development

Flux Balance Analysis (FBA) is a mathematical approach used to analyze the flow of metabolites within a metabolic network [2]. Often employed alongside genome-scale metabolic models (GEMs), FBA helps assess and optimize biochemical reactions in complex biological systems. GEMs serve as comprehensive frameworks for contextualizing microbial omics studies, such as metagenomics, metatranscriptomics, and metabolomics, by providing a structured representation of metabolic pathways [3].

In simpler terms, GEMs establish a fixed metabolic pathway, or "chassis," while methods like FBA examine specific reactions of interest. In the context of FBA, "flux" refers to the rate at which material is processed or converted by a biochemical reaction, indicating the speed at which a given reaction occurs within the network.

FBA interprets the system’s stability by ensuring that metabolite production matches consumption rates, aiming to optimize flux distribution across the network. This optimization is often done to maximize biomass production. In our case, we are using FBA to optimize holdfast production and determine the ideal reaction conditions for future scale-up efforts.

We expanded the BL21(DE3) model (iHK1487) developed by Hanseol Kim et al. (2018) by implementing the HfsE/J/G/L/H pathway, which synthesizes the precursor to holdfast [4]. In total, we added 5 new reactions and 5 metabolites (intermediate products of the HfsE/J/G/L/H pathway):

HfsE

Undecaprenyl pyrophosphate + UDP-Glucose -> Intermediate 1

HfsJ

Intermediate 1 + UDP-Mannosamine uronic acid -> Intermediate 2

HfsG

Intermediate 2 + UDP-N-Acetyl-Glucosamine -> Intermediate 3

HfsL

Intermediate 3 + UDP-N-Acteyl-Glucosamine -> Intermediate 4

HfsH

Intermediate 4 -> Holdfast precursor + Acetate

As a validation for our model, we performed FBA while targeting the HsfH as the objective reaction, with 8 different carbon sources (medium exchange reaction capped at 10 mmol gDCWsup^-1 h^-1) in the medium. Additionally, we ran a dot blot analysis of intact ring cells from flasks with different carbon sources as a ground truth against which to compare the model results.

These results encouraged us, as the model predicted the holdfast production performance fairly accurately. The only exception was sucrose, as the model repeatedly failed to find an optimal solution, and all troubleshooting attempts failed. (We suspect it might have been an issue with the PEP:PYR sucrose import system reactions not being configured properly.) Additionally, glucose was confirmed to be the best-performing carbon source; therefore, all further analyses utilized it as the primary carbon source.

Results

Based on the consumption/production rate predictions, we made a couple of observations on the optimal amount of substrates to be supplied to the growth medium:

• Glucose: no additional benefits were observed when supplied more than 20 mmol gDCW^-1 h^-1;
• Ammonia: no additional benefits were observed when supplied more than 8 mmol gDCW^-1 h^-1;
• Carbon dioxide: no additional carbon dioxide is necessary for holdfast production;
• Oxygen: no additional benefits were observed above when supplied more than 13 mmol gDCW^-1 h^-1;

Conclusion

We successfully extended the iHK1487 model from the literature with our holdfast precursor production pathway [4]. The model was validated by comparing the predicted fluxes against a dot blot assay of the intact ring cells grown on different carbon sources. We used the validated model to find the optimal amount of substrates to be supplied to the growth medium. These predictions will serve as a crucial starting point for efficiently scaling up holdfast production, should we move forward with mass production of the adhesive.

Key References

1. Ebrahim, A., Lerman, J. A., Palsson, B. O., & Hyduke, D. R. (2013). COBRApy: COnstraints-Based Reconstruction and Analysis for Python.BMC Systems Biology, 7(1), 74. doi: https://doi.org/10.1186/1752-0509-7-74.
2. Orth, J. D., Thiele, I., & Palsson, B. Ø. (2010). What is flux balance analysis?Nature Biotechnology, 28(3), 245–248. doi: https://doi.org/10.1038/nbt.1614.
3. Passi, A., Tibocha-Bonilla, J. D., Kumar, M., Tec-Campos, D., Zengler, K., & Zuniga, C. (2022). Genome-Scale Metabolic Modeling Enables In-Depth Understanding of Big Data.Metabolites, 12(1), 14. doi: https://doi.org/10.3390/metabo12010014.
4. Kim, H., Kim, S., & Sung Ho Yoon. (2018). Metabolic network reconstruction and phenome analysis of the industrial microbe, Escherichia coli BL21(DE3).PloS One, 13(9), e0204375–e0204375. doi: https://doi.org/10.1371/journal.pone.0204375.
5. King, Z. A., Dräger, A., Ebrahim, A., Sonnenschein, N., Lewis, N. E., & Palsson, B. O. (2015). Escher: A Web Application for Building, Sharing, and Embedding Data-Rich Visualizations of Biological Pathways.PLOS Computational Biology, 11(8), e1004321. doi: https://doi.org/10.1371/journal.pcbi.1004321.