On the scope and limitations of baker’s yeast as a model organism for studying human tissue-specific pathways

Shahin Mohammadi, Baharak Saberidokht, Shankar Subramaniam, Ananth GramabioRxiv 2014 · DOI

TL;DR

This paper introduces a computational framework to assess how well baker's yeast (S. cerevisiae) can model human tissue-specific biological pathways. By analyzing conserved and human-specific gene networks, they identify which human tissues are best represented by yeast and highlight the potential of human-specific genes for drug target discovery.

What Problem Does This Paper Address?

Baker's yeast has been a workhorse for biological research due to its ease of use and genetic tractability. However, humans are complex multicellular organisms with highly specialized tissues, each with unique physiological properties. While many fundamental cellular processes are conserved, it's unclear to what extent yeast can accurately model the distinct pathways that govern specific human tissues and contribute to tissue-specific diseases. This paper aims to bridge this gap by systematically quantifying the suitability of yeast as a model for different human tissues.

Methods

The authors developed a novel computational and statistical framework. Key steps include:

  • Human Interactome Dissection: They first created tissue-specific human cellular networks by integrating protein-protein interaction (PPI) data with gene expression data from various human tissues.
  • Network Alignment: A sparse network alignment method based on belief propagation was used to align these human tissue-specific networks with the yeast interactome. This method considers both protein sequence similarity and the topology of interactions.
  • Gene Partitioning: Genes were then partitioned into "housekeeping" (universally expressed) and "tissue-selective" (expressed in specific tissues) categories. These categories were further divided into "conserved" (shared with yeast) and "human-specific" (unique to humans).
  • Tissue Similarity Modeling: A statistical model (Tissue-Specific Random Model - TRAM) was developed to quantify the similarity between each human tissue and the yeast interactome, considering both gene expression and interaction network topology.
  • Functional Enrichment Analysis: Tools like g:Profiler and EnrichmentMap were used to identify the functional roles of conserved and human-specific gene sets within different tissue groups.

Datasets

  • Human Protein-Protein Interaction (PPI) Networks: Integrated from 21 databases.
  • Human Gene Expression Data: From the GNF Gene Atlas.
  • Yeast Interactome: Obtained from the BioGRID database.
  • Protein Sequences: Downloaded from Ensembl.
  • Genetic Association Database (GAD): Used for disease enrichment analysis.

Key Insights and Ideas

  • Yeast as a Model Varies by Tissue: The study reveals that yeast's suitability as a model organism is not uniform across all human tissues. Some tissues, like blood cells and male reproductive tissues, show higher similarity to yeast, while others, like ganglion tissues, show less.
  • Housekeeping vs. Tissue-Selective Genes: A clear distinction emerged between conserved housekeeping genes (fundamental cellular processes, often conserved with yeast) and human-specific housekeeping genes (involved in cell-to-cell communication, anatomical development).
  • Human-Specific Genes as Disease Drivers: Human-specific tissue-selective genes were found to be significantly associated with tissue-specific pathologies, suggesting they are key drivers of disease in those tissues.
  • Potential for Targeted Therapies: The identification of human-specific genes linked to diseases points to them as promising candidates for novel drug targets and for developing more accurate humanized yeast models for studying these diseases.

Key Results

  • The framework successfully partitioned human genes into conserved and human-specific categories across different tissue types.
  • Blood cells and testis tissues showed the highest similarity to the yeast interactome, while ganglion tissues showed the least.
  • Conserved housekeeping genes were enriched in core cellular functions like ribosome biogenesis and translation, while human-specific housekeeping genes were linked to cell communication and development.
  • Human-specific tissue-selective genes were strongly correlated with tissue-specific pathologies, particularly in brain and blood cells. For example, human-specific brain genes were better predictors of psychiatric disorders than conserved genes.

Limitations and Future Work

  • Artifacts in Human-Specific Gene Classification: The authors acknowledge that some genes classified as human-specific might be artifacts of the method, particularly those with no sequence similarity to yeast genes. These genes might also represent under-annotated regions of the human genome.
  • Unicellular vs. Multicellular: As a unicellular organism, yeast cannot fully capture organism-level phenotypes that arise from inter-cellular interactions.
  • Future Directions: The study suggests that humanized yeast models, created by introducing specific human genes, can be valuable for studying tissue-specific physiology and pathophysiology. The identified human-specific genes can guide the design of these models and serve as targets for therapeutic interventions. Further validation with more physiologically relevant model organisms is also suggested.

Summary updated

Abstract

Abstract Budding yeast, S. cerevisiae, has been used extensively as a model organism for studying cellular processes in evolutionarily distant species, including humans. However, different human tissues, while inheriting a similar genetic code, exhibit distinct anatomical and physiological properties. Specific biochemical processes and associated biomolecules that differentiate various tissues are not completely understood, neither is the extent to which a unicellular organism, such as yeast, can be used to model these processes within each tissue. We propose a novel computational and statistical framework to systematically quantify the suitability of yeast as a model organism for different human tissues. We develop a computational method for dissecting the human interactome into tissue-specific cellular networks. Using these networks, we simultaneously partition the functional space of human genes, and their corresponding pathways, based on their conservation both across species and among different tissues. We study these subspaces in detail, and relate them to the overall similarity of each tissue with yeast. Many complex disorders are driven by a coupling of housekeeping (universally expressed in all tissues) and tissue-selective (expressed only in specific tissues) dysregulated pathways. We show that human-specific subsets of tissue-selective genes are significantly associated with the onset and development of a number of pathologies. Consequently, they provide excellent candidates as drug targets for therapeutic interventions. We also present a novel tool that can be used to assess the suitability of the yeast model for studying tissue-specific physiology and pathophysiology in humans.
BibTeX
@misc{mohammadi2014scope,
  author = {Shahin Mohammadi and Baharak Saberidokht and Shankar Subramaniam and Ananth Grama},
  title = {{On the scope and limitations of baker’s yeast as a model organism for studying human tissue-specific pathways}},
  year = {2014},
  doi = {10.1101/011858},
  howpublished = {bioRxiv},
}
Figure