The protein structurome of Orthornavirae and its dark matter

Pascal Mutz, Antônio Pedro Camargo, Harutyun Sahakyan, Uri Neri, Anamarija Butković, Yuri I. Wolf, Mart Krupovìč, Valerian V. Dolja, Eugene V. KooninmBio 2024 · DOI

TL;DR

Researchers have built the first comprehensive "structurome" of RNA viruses (Orthornavirae) by predicting protein structures for their uncharacterized "dark matter" proteins. This reveals that most of this dark matter consists of disordered or simple helical structures, likely involved in virus-host interactions, while the majority of conserved globular domains are already known.

What Problem Does This Paper Address?

RNA viruses, particularly those in the Orthornavirae kingdom, are incredibly diverse and rapidly evolving. While many viral proteins have known functions, a significant portion remains uncharacterized – the "dark matter" of their proteomes. This dark matter is challenging to analyze using traditional sequence-based methods due to rapid evolution. Understanding these unknown proteins is crucial for deciphering viral functions, evolution, and host interactions. The authors aim to shed light on this dark matter by leveraging advanced protein structure prediction tools.

Methods

The study employed a multi-pronged approach:

  • Data Acquisition: They utilized a large dataset of predicted RNA virus genomes from metatranscriptomics.
  • Protein and Domain Identification: Standard methods were used to identify annotated proteins and extract unannotated regions (conserved unannotated domains - CUDs) and hypothetical open reading frames (ORFans).
  • Structure Prediction:
    • AlphaFold2: Used for modeling proteins and CUDs with high confidence.
  • ESM-Fold: Employed for modeling ORFans due to computational cost constraints of AlphaFold2.
  • Structure Comparison:
    • Foldseek: Used for initial pre-clustering of predicted structures.
  • Dali: Utilized for all-vs-all structure comparisons to identify similarities and relationships.
  • "Structurome" Construction: A comprehensive database of predicted viral protein structures was built.
  • Functional Annotation: Predicted structures were compared against known protein databases (PDB) and analyzed for secondary structure and potential functional domains.
  • Neighborhood Analysis: A method to infer function by examining homologous proteins in neighboring genomes.

Datasets

  • EMRV set: A dataset of predicted orthornavirus genomes from metatranscriptomics (over 370,000 viral contigs from nearly 500 virus families).
  • ICTV Exemplars: A curated set of viral reference genomes from the International Committee on Taxonomy of Viruses (ICTV) for 98 virus families.
  • PDB: The Protein Data Bank, used for structural comparisons.

Key Insights and Ideas

  • The "Dark Matter" Revealed: The study successfully characterized a significant portion of the "dark matter" in RNA virus proteomes, showing it's not just random noise but often consists of intrinsically disordered or simple all-α-helical structures.
  • Disordered Domains Dominate: A large fraction of the unannotated regions and ORFans predicted to have low confidence models were found to be disordered or simple, suggesting they play roles in dynamic interactions rather than stable structural functions.
  • Known Globular Domains are Largely Identified: The research indicates that the vast majority of conserved, globular protein domains in RNA viruses have likely already been identified. New discoveries tend to be lineage-specific.
  • Exaptation is a Key Evolutionary Driver: The study identified a compelling case of exaptation where a cellular nucleoside monophosphate kinase was repurposed as an RNA-binding protein in several virus families, highlighting the dynamic nature of viral evolution.
  • Structure Prediction is Crucial for Viral Evolution: The rapid evolution of viruses makes sequence-based analysis insufficient. Protein structure prediction, as demonstrated here, is essential for inferring function and evolutionary relationships.

Key Results

  • The constructed "structurome" comprises thousands of predicted protein structures from orthornaviruses.
  • Approximately 34% of CUDs and 10% of ORFans could be predicted with high confidence (mean plddt score ≥70).
  • A significant portion of the "dark matter" (CUDs and ORFans) was predicted to have either generic folds (like single α-helices) or no high-confidence structure, suggesting intrinsically disordered regions.
  • Several lineage-specific globular domains, including nucleic acid-binding domains and nucleases, were identified within the proteomic dark matter.
  • A specific example of exaptation was found: a cellular nucleoside monophosphate kinase was repurposed as an RNA-binding protein in multiple virus families.

Limitations and Future Work

  • Low-Confidence Models: A substantial fraction of predicted structures yielded low confidence scores, indicating intrinsically disordered regions or challenging modeling scenarios.
  • "Generic" Folds: Many unannotated regions resulted in simple or generic folds that are difficult to assign specific functions to.
  • Lineage-Specific Discoveries: While new domains were found, they were largely confined to specific virus families, suggesting that broadly conserved novel domains are rare.
  • Future Work: The authors suggest that further experimental studies are needed to uncover the precise functions of these newly identified domains and disordered regions. The ongoing discovery of new viruses through metatranscriptomics will continue to benefit from advanced protein structure modeling and comparison techniques.

Summary updated

Abstract

Advanced methods for protein structure prediction, such as AlphaFold2, greatly expand our capability to identify protein domains and infer their likely functions and evolutionary relationships. This is particularly pertinent for proteins encoded by viruses that are known to evolve rapidly and as a result often cannot be adequately characterized by analysis of the protein sequences. We performed an exhaustive structure prediction and comparative analysis for uncharacterized proteins and domains ("dark matter") encoded by viruses with RNA genomes. The results show the dark matter of RNA virus proteome consists mostly of disordered and all-α-helical domains that cannot be readily assigned a specific function and that likely mediate various interactions between viral proteins and between viral and host proteins. The great majority of globular proteins and domains of RNA viruses are already known although we identified several unexpected domains represented in individual viral families.
BibTeX
@inproceedings{mutz2024protein,
  author = {Pascal Mutz and Antônio Pedro Camargo and Harutyun Sahakyan and Uri Neri and Anamarija Butković and Yuri I. Wolf and Mart Krupovìč and Valerian V. Dolja and Eugene V. Koonin},
  title = {{The protein structurome of Orthornavirae and its dark matter}},
  booktitle = {mBio},
  year = {2024},
  doi = {10.1128/mbio.03200-24},
}
Figure