Conserved degronome features governing quality control associated proteolysis

Bayan Mashahreh, Shir Armony, Kristoffer E. Johansson, Alon Chappleboim, Nir Friedman, Richard G. Gardner, Rasmus Hartmann‐Petersen, Kresten Lindorff‐Larsen, Tommer Ravid

TL;DR

Researchers have developed a novel yeast-based system to identify protein degradation signals (degrons) and used machine learning to uncover key features of these signals. They discovered that transmembrane domain-like sequences are particularly potent degrons, suggesting a conserved mechanism for protein quality control across evolution.

What Problem Does This Paper Address?

Maintaining protein homeostasis is crucial for cellular health. Eukaryotic cells employ sophisticated quality control (QC) systems to identify and eliminate misfolded or damaged proteins. A key mechanism involves the ubiquitin-proteasome system (UPS), which tags proteins for degradation. This tagging process relies on the recognition of specific degradation determinants, called "degrons," by E3 ubiquitin-protein ligases. While many degrons have been identified, they often arise from studies of regulated degradation (like cell division or cancer), and the full repertoire of degrons, especially those involved in general protein QC, remains largely unknown. This gap limits our understanding of how cells manage the vast diversity of protein half-lives and respond to cellular stress. The authors aim to systematically identify and characterize these quality control degrons.

Methods

The core of this study is the development and application of a yeast-adapted Global Protein Stability (yGPS-P) peptidome technology. This system builds upon a fluorescence-based reporter assay. Here's how it works:

  1. Reporter System: A bicistronic gene expression system is used, producing both yeast-enhanced Cherry (yeC) and yeast-enhanced Green Fluorescent Protein (yeG) from a single transcript. Crucially, a peptide sequence is fused to the C-terminus of yeG.
  2. Peptide Library: A library of short peptide sequences (51-mers with 36-mer overlaps) derived from yeast proteins is generated and cloned into the yGPS-P vector. This creates a diverse collection of yeast proteins fused to yeG.
  3. Stability Measurement: When these fusion proteins are expressed in yeast, the ratio of yeG to yeC fluorescence (yeG/yeC) serves as a proxy for the stability of the fused peptide. A lower yeG/yeC ratio indicates that the fused peptide is acting as a degron, targeting the yeG fusion protein for degradation by the UPS.
  4. Machine Learning for Prediction: After identifying peptides that significantly decrease yeG/yeC, a machine-learning algorithm called QCDPred was developed. This algorithm was trained on the amino acid composition of these identified degrons to predict the probability of any given sequence acting as a degron.

This approach allows for an unbiased screen of a large portion of the proteome to identify novel degrons and then to learn the underlying sequence features that confer degron activity.

Datasets

  • Yeast Proteome-Derived Peptidome Library: The study generated a library of approximately 29,500 DNA fragments (51-mers with 12-amino acid overlaps) derived from 326 yeast proteins. These proteins were selected as components of multi-protein complexes.
  • Human Proteome Data: For cross-species comparison, data from a previous study by Koren et al. on C-terminal regions of the human proteome was used.
  • Viral Protein Data: Influenza C virus p42 protein was used as a test case for predicting degrons in a viral context.

Key Insights and Ideas

  • Transmembrane Domains (TMDs) as Potent Degrons: Perhaps the most striking finding is that sequences resembling transmembrane domains (TMDs) are exceptionally effective degrons. This was unexpected, as the machine learning model was not explicitly trained to recognize TMDs, but rather learned this feature from the data.
  • Hydrophobicity is Key: The study reinforces the idea that hydrophobicity is a critical feature of degrons, aligning with the nature of TMDs.
  • Conserved Degradation Mechanisms: The identification of similar degron features in viral and human proteins strongly suggests that these quality control mechanisms are evolutionarily conserved.
  • Degrons are Often "Cryptic": The data implies that many degrons are not always active but may become exposed or functional when a protein is perturbed or misfolded, suggesting a dynamic role in protein homeostasis.
  • Negative Charge is a Degron Inhibitor: The presence of negatively charged amino acids (glutamate and aspartate) significantly reduces a sequence's ability to act as a degron.
  • Sequence vs. Chemical Properties: While the exact linear sequence of a degron can modulate its strength, the general chemical properties (like hydrophobicity) appear to be more dominant determinants of degron activity.

Key Results

  • Identification of a Large Cohort of Degrons: The yGPS-P screen identified a significant number of proteome-derived degrons in yeast.
  • QCDPred Model Performance: The developed QCDPred algorithm showed good predictive power, with experimental validation confirming its ability to identify functional degrons.
  • TMDs Show Highest Potency: Degrons with TMD-like characteristics consistently received the highest probability scores from QCDPred, indicating their strong degron activity.
  • Cross-Species Conservation: The principles of degron prediction and features identified in yeast were found to correlate with human proteome data, suggesting evolutionary conservation.
  • Functional Enrichment: Pathways related to transport and lipid metabolism were enriched in proteins with high degron probability, aligning with the prevalence of membrane-associated proteins. Conversely, pathways involved in transcription, translation, and protein folding were depleted of degrons, suggesting a mechanism to protect essential QC machinery.

Limitations and Future Work

  • Context Dependency: The authors acknowledge that the QCDPred algorithm, trained on short peptides, is devoid of cellular context. While it predicts degron potential, the actual function in a full protein can be influenced by factors like protein folding and localization.
  • C-terminal Degrons in Yeast: Unlike in humans, C-terminal degrons did not appear to play a significant role in yeast protein turnover, suggesting differences in degradation machinery or recognition mechanisms between species. Further studies are needed to explore this discrepancy.
  • E3 Ligase Specificity: While the study identified degrons, further work is needed to fully elucidate the specificity of different E3 ligases for these degrons and how this specificity contributes to the overall protein QC landscape.
  • Broader Proteome and Organismal Scope: The authors suggest extending the QCDPred tool to analyze the degronome of other organisms and protein types, including those involved in disease. They also propose further investigation into the role of chaperones in mediating degron recognition.

Summary updated

Abstract

The eukaryotic proteome undergoes constant surveillance by quality control systems that either sequester, refold, or eliminate aberrant proteins by ubiquitin-dependent mechanisms. Ubiquitin-conjugation necessitates the recognition of degradation determinants, termed degrons, by their cognate E3 ubiquitin-protein ligases. To learn about the distinctive properties of quality control degrons, we performed an unbiased peptidome stability screen in yeast. The search identify a large cohort of proteome-derived degrons, some of which exhibited broad E3 ligase specificity. Consequent application of a machine-learning algorithm establishes constraints governing degron potency, including the amino acid composition and secondary structure propensities. According to the set criteria, degrons with transmembrane domain-like characteristics are the most probable sequences to act as degrons. Similar quality control degrons are present in viral and human proteins, suggesting conserved degradation mechanisms. Altogether, the emerging data indicate that transmembrane domain-like degron features have been preserved in evolution as key quality control determinants of protein half-life.
BibTeX
@inproceedings{mashahreh2022conserved,
  author = {Bayan Mashahreh and Shir Armony and Kristoffer E. Johansson and Alon Chappleboim and Nir Friedman and Richard G. Gardner and Rasmus Hartmann‐Petersen and Kresten Lindorff‐Larsen and Tommer Ravid},
  title = {{Conserved degronome features governing quality control associated proteolysis}},
  booktitle = {Nature Communications},
  year = {2022},
  doi = {10.1038/s41467-022-35298-y},
}
Figure