Domain Rearrangements in Protein Evolution

Åsa K. Björklund, Diana Ekman, Sara Light, Johannes Frey-Skött, Arne ElofssonJournal of Molecular Biology 2005 · DOI

TL;DR

This paper introduces "domain distance," a novel metric to quantify evolutionary changes in protein domain architectures. By applying this metric, the authors reveal that domain insertions and deletions (indels) are the primary drivers of multi-domain protein evolution, far more common than domain exchanges, and that these events predominantly occur at protein termini.

What Problem Does This Paper Address?

Proteins are the workhorses of biology, and their diverse functions often arise from combinations of modular protein domains. Understanding how these multi-domain proteins evolve – how new combinations arise and how existing ones change – is crucial for deciphering evolutionary history and predicting protein function. Traditional methods for studying protein evolution often rely on sequence similarity, which can be unreliable for distantly related proteins or those with complex domain arrangements. The challenge is to find a robust way to compare and contrast the "domain architectures" of proteins to infer their evolutionary relationships and the events that shaped them.

Methods

The core innovation of this paper is the domain distance (DD) metric. DD quantifies the difference between two protein domain architectures by counting the number of unmatched domains in an alignment of their constituent domains. This approach treats each domain family as a symbol, allowing for comparisons even when sequences are highly divergent.

The authors also employed:

  • Sequence alignment: Standard methods like ClustalW were used for comparison with domain-based approaches.
  • Gene Ontology (GO) annotations: Semantic similarity derived from GO terms was used to assess functional relatedness and compare it with domain distance.
  • Phylogenetic tree construction: Evolutionary trees were built using both sequence alignments and the novel domain distance metric to visualize evolutionary relationships.

Datasets

The study utilized several large-scale biological datasets:

  • SWISS-PROT release 44: A comprehensive protein sequence database.
  • Pfam-A and Pfam-B: Databases of protein domain families.
  • Seven eukaryotic proteomes: Including human, mouse, fly, worm, yeast, and plants.
  • SCOP: A structural classification of proteins database, used for comparison with Pfam.

The authors also created a non-redundant subset of SWISS-PROT proteins with GO annotations for functional similarity analysis.

Key Insights and Ideas

  • Domain distance as a robust evolutionary metric: DD provides a powerful way to compare protein architectures, overcoming limitations of sequence similarity for distantly related proteins. It's less sensitive to sequence divergence within domains.
  • Indels dominate protein evolution: The study strongly suggests that the addition (insertion) and removal (deletion) of entire domains are the most frequent evolutionary events shaping multi-domain proteins.
  • Termini are hot spots for evolution: These domain insertions and deletions predominantly occur at the N- and C-termini of proteins, rather than in the middle of existing domain arrangements.
  • Domain exchanges are rare: The substitution of one domain for another is a much less common evolutionary event.
  • Functional divergence through domain addition: Adding new domains to a protein architecture is a key mechanism for acquiring new functions or fine-tuning existing ones, particularly for binding activities.
  • Evolutionary trees can be built from domain architectures: The domain distance metric enables the construction of phylogenetic trees that can reveal evolutionary relationships and events, sometimes offering different insights than sequence-based trees (e.g., clustering of PH-domain adaptors).

Key Results

  • Indels are ~3-4 times more frequent than repetitions, and domain exchanges are significantly rarer (making up only ~1-5% of events across datasets).
  • N-terminal indels show a slight preference, while C-terminal exchanges are more common (though rare overall). Repeats are evenly distributed at both ends.
  • Domain distance correlates well with sequence similarity and semantic similarity based on GO annotations, validating DD as a meaningful measure of evolutionary and functional relatedness.
  • For many protein families (e.g., SH2-containing kinases), domain distance trees provide a clear evolutionary scenario, highlighting the role of domain additions in shaping protein function and specificity. For instance, the evolution of kinases often involves N-terminal insertions of regulatory or binding domains.
  • Domain families involved in binding functions are more likely to be added to existing architectures, while enzymatic domains are less frequently added.

Limitations and Future Work

  • Domain assignment imperfections: The accuracy of the analysis is dependent on the quality and completeness of domain assignments in databases like Pfam. Unassigned regions or misclassified domains can introduce noise.
  • Ambiguity in evolutionary events: Distinguishing between insertions and deletions can be challenging solely from domain architectures. The assumption that evolution generally leads to more complex architectures (insertions/fusions) is made.
  • Potential for novel mechanisms: The authors acknowledge that other evolutionary mechanisms might exist and that their model is based on the assumption that the closest neighbor in domain distance represents the direct ancestor.
  • Future directions: The authors suggest that a more elaborate definition of domain distances, potentially incorporating independent weights for different domain families or distinguishing between repeating and non-repeating regions, could further refine the analysis. Investigating the evolutionary constraints on different domain combinations is also a promising avenue.

Summary updated

BibTeX
@article{bjorklund2005domain,
  author = {Åsa K. Björklund and Diana Ekman and Sara Light and Johannes Frey-Skött and Arne Elofsson},
  title = {{Domain Rearrangements in Protein Evolution}},
  journal = {Journal of Molecular Biology},
  year = {2005},
  doi = {10.1016/j.jmb.2005.08.067},
}

Loading related content...

Figure