Multi-Omics Driven Metabolic Network Reconstruction and Analysis of Lignocellulosic Carbon Utilization in Rhodosporidium toruloides

Joonhoon Kim, Samuel T. Coradetti, Young‐Mo Kim, Yuqian Gao, Junko Yaegashi, Jeremy Zucker, Nathalie Munoz Munoz, Erika Zink, Kristin Burnum-Johnson, Scott Baker, Blake A. Simmons, Jeffrey M. Skerker, John M. Gladden, Jon MagnusonFrontiers in Bioengineering and Biotechnology 2021 · DOI

TL;DR

Researchers have developed the most comprehensive genome-scale metabolic network model to date for Rhodosporidium toruloides, a yeast capable of converting plant biomass into valuable products. By integrating multi-omics data (genomics, transcriptomics, proteomics, metabolomics) and rigorous manual curation, they've created a powerful tool to understand and engineer this organism for biofuel and bioproduct production.

What Problem Does This Paper Address?

Rhodosporidium toruloides is a highly promising "non-model" organism for industrial biotechnology. It's an oleaginous yeast, meaning it can accumulate significant amounts of lipids, making it a prime candidate for producing biofuels and other valuable chemicals from lignocellulosic biomass – the tough, fibrous material found in plants. However, our understanding of its metabolic capabilities, especially how it breaks down and utilizes the complex sugars and aromatic compounds found in plant biomass hydrolysates, is limited. To effectively engineer R. toruloides for specific applications, we need a detailed, accurate map of its metabolic pathways. This paper aims to fill that gap by reconstructing and validating a genome-scale metabolic network model for R. toruloides.

Methods

The core of this work is the meticulous reconstruction and curation of a genome-scale metabolic network. Here's a breakdown of their approach:

  • Draft Reconstruction: They started by leveraging existing, high-quality metabolic network models of well-studied organisms (like Saccharomyces cerevisiae and E. coli) and mapped orthologous proteins from R. toruloides to these models. This provided a foundational draft network.
  • Manual Curation: This was a critical step. The draft network was manually refined using a wealth of data:
  • Functional Annotation: Detailed examination of gene and protein functions.
  • Multi-Omics Data: Integrating transcriptomics (gene expression), proteomics (protein abundance), metabolomics (metabolite levels), and RB-TDNAseq (gene fitness data). This allowed them to identify active pathways and essential genes under various conditions.
  • Localization Prediction: Determining where proteins function within the cell (cytosol, mitochondria, peroxisomes, etc.) is crucial for accurate pathway reconstruction.
  • Literature Mining: Incorporating known metabolic pathways and experimental findings from previous studies.
  • Model Building and Evaluation: The curated network was then used to build a functional metabolic model. This involved:
  • Biomass Equation: Defining the cellular components needed for growth.
  • Transport Reactions: Modeling how substances move in and out of the cell and its compartments.
  • Growth Simulations: Using Flux Balance Analysis (FBA) to predict growth rates and identify bottlenecks.
  • Validation: The model's predictions were rigorously tested against experimental data:
  • Biolog Phenotype Microarrays: High-throughput experiments measuring growth on a wide range of carbon, nitrogen, sulfur, and phosphate sources.
  • RB-TDNAseq: This technique measures the fitness (growth impact) of thousands of random gene insertion mutants, providing insights into gene essentiality under specific conditions.

Datasets

The study utilized several key datasets:

  • Genome Annotation: From a previous study of R. toruloides IFO0880.
  • Orthologous Protein Data: Generated using OrthoMCL to identify related proteins across different species.
  • Existing Metabolic Models: From databases like BiGG Models, and specific models of S. cerevisiae, C. reinhardtii, human, mouse, E. coli, P. putida, and Y. lipolytica.
  • Multi-Omics Data (Newly Generated/Analyzed):
    • Transcriptomics: RNA sequencing data from R. toruloides grown on various carbon sources (glucose, D-xylose, L-arabinose, p-coumarate).
  • Proteomics: Protein abundance data from the same growth conditions.
  • Metabolomics: Measurement of intracellular and extracellular metabolites.
  • RB-TDNAseq: Gene fitness data under various conditions.
  • Phenotype Microarray Data: From Biolog plates for growth phenotype validation.

Key Insights and Ideas

  • An Alternative Pentose Utilization Pathway: The study suggests R. toruloides might use a different route to break down pentose sugars (like D-xylose and L-arabinose) than previously thought. Instead of going through D-xylulose-5-phosphate, it appears to involve D-arabinitol and D-ribulose, eventually feeding into the pentose phosphate pathway. This is supported by multi-omics data and the lack of fitness defects in key enzymes of the proposed alternative pathway.
  • Peroxisomal Beta-Oxidation for Aromatic Compound Degradation: For p-coumarate, a common component of plant biomass, the researchers propose a pathway involving peroxisomal beta-oxidation-like enzymes, leading to protocatechuate. This intermediate is then further processed in the cytosol and mitochondria. This pathway is distinct from typical mitochondrial or peroxisomal fatty acid beta-oxidation.
  • Importance of Compartmentalization: Accurately assigning reactions and metabolites to specific cellular compartments (like mitochondria and peroxisomes) was crucial for refining the model and understanding pathway localization.
  • Lipid Body as a Separate Entity: The model explicitly separates the lipid body from the main biomass equation. This allows for independent simulation of lipid accumulation and mobilization, which is key for an oleaginous yeast.
  • Multi-Omics Integration is Key: The success of this reconstruction hinges on the synergistic use of transcriptomics, proteomics, metabolomics, and gene fitness data. Each data type provided complementary information to build and validate the model.

Key Results

  • Most Complete Metabolic Model: The developed model is described as the "most complete and accurate metabolic network model available for R. toruloides to date."
  • Model Size: The final curated model includes 1141 genes, 2398 reactions, and 2051 metabolites distributed across nine compartments.
  • Prediction Accuracy:
    • The model accurately predicted 78.4% of growth phenotypes from Biolog plates after manual curation.
  • It achieved 78.6% accuracy in predicting gene essentiality across 27 conditions.
  • Identification of Novel Pathways: The study identified potential novel pathways for pentose and aromatic compound utilization, supported by multi-omics evidence.

Limitations and Future Work

  • Ortholog Identification Risks: The authors acknowledge the potential for incorrect reaction identification due to false positives in ortholog mapping, especially considering the possibility of horizontal gene transfer (HGT) in yeasts. While they found no direct evidence of large-scale HGT in R. toruloides, they note that future models will need to account for this.
  • Experimental Validation Needed: While multi-omics data strongly supports the proposed novel pathways, direct experimental validation of these specific pathways is still needed.
  • Further Refinement: The authors suggest that the model will require ongoing improvement and re-evaluation as new metabolic pathways are characterized and as more data becomes available.
  • Broader Applications: The developed model and multi-omics data are presented as a valuable resource for future studies on R. toruloides, particularly for engineering it for lignocellulosic biomass conversion into biofuels and bioproducts.

Summary updated

Abstract

An oleaginous yeast <i>Rhodosporidium toruloides</i> is a promising host for converting lignocellulosic biomass to bioproducts and biofuels. In this work, we performed multi-omics analysis of lignocellulosic carbon utilization in <i>R. toruloides</i> and reconstructed the genome-scale metabolic network of <i>R. toruloides</i>. High-quality metabolic network models for model organisms and orthologous protein mapping were used to build a draft metabolic network reconstruction. The reconstruction was manually curated to build a metabolic model using functional annotation and multi-omics data including transcriptomics, proteomics, metabolomics, and RB-TDNA sequencing. The multi-omics data and metabolic model were used to investigate <i>R. toruloides</i> metabolism including lipid accumulation and lignocellulosic carbon utilization. The developed metabolic model was validated against high-throughput growth phenotyping and gene fitness data, and further refined to resolve the inconsistencies between prediction and data. We believe that this is the most complete and accurate metabolic network model available for <i>R. toruloides</i> to date.
BibTeX
@inproceedings{kim2021multi,
  author = {Joonhoon Kim and Samuel T. Coradetti and Young‐Mo Kim and Yuqian Gao and Junko Yaegashi and Jeremy Zucker and Nathalie Munoz Munoz and Erika Zink and Kristin Burnum-Johnson and Scott Baker and Blake A. Simmons and Jeffrey M. Skerker and John M. Gladden and Jon Magnuson},
  title = {{Multi-Omics Driven Metabolic Network Reconstruction and Analysis of Lignocellulosic Carbon Utilization in Rhodosporidium toruloides}},
  booktitle = {Frontiers in Bioengineering and Biotechnology},
  year = {2021},
  doi = {10.3389/fbioe.2020.612832},
}
Figure