Surveying phylogenetic footprints in large gene clusters: applications to Hox cluster duplications

Sonja J. Prohaska, Claudia Fried, Christoph Flamm, Günter P. Wagner, Peter F. StadlerMolecular 2003 · DOI

TL;DR

This paper introduces a new computational method, "tracker," for identifying conserved non-coding DNA regions (phylogenetic footprints) in large gene clusters across multiple species. The authors apply this tool to Hox gene clusters, revealing significant loss of sequence conservation following gene duplication events, suggesting that evolutionary pressures beyond simple gene loss are at play.

What Problem Does This Paper Address?

Identifying regulatory elements in non-coding DNA is crucial for understanding gene regulation and evolution. Phylogenetic footprints, regions of conserved non-coding DNA across species, are a promising source for discovering these elements. However, existing methods often struggle with large gene clusters and complex evolutionary histories, such as gene duplication. This paper addresses the challenge of efficiently and accurately surveying these footprints in large genomic regions, particularly in the context of Hox gene cluster duplications, to understand their evolutionary dynamics.

Methods

The core of this work is the "tracker" software. It employs a multi-step approach:

  1. Pairwise Alignments: It begins with a sensitive, but potentially noisy, pairwise alignment of all input sequences using blastz.
  2. Low Complexity Filtering: A local entropy criterion is used to identify and remove repetitive or low-complexity regions that can lead to spurious alignments.
  3. Alignment Assembly and Filtering: Alignments are then re-aligned using a standard algorithm like clustalw and filtered based on identity and quality thresholds.
  4. Consistent Cliques: The method identifies "consistent cliques" from overlapping alignments, representing potentially conserved footprints. This involves constructing an incompatibility graph and finding its maximal cliques.
  5. Multiple Alignments: For each identified clique, multiple alignments are generated to visualize conserved regions.
  6. Phylogenetic Distribution: Finally, the presence/absence patterns of footprints are mapped onto a phylogenetic tree to study their evolutionary distribution.

The key novelty lies in the systematic approach to handling large clusters, the clique-finding algorithm for identifying consistent footprints, and the application to complex evolutionary scenarios like gene duplication.

Datasets

The study utilizes publicly available genomic sequences for Hox gene clusters from:

  • Human (Homo sapiens)
  • Shark (Heterodontus francisci)
  • Zebrafish (Danio rerio)
  • Pufferfish (Takifugu rubripes)

The paper also refers to existing databases and previous studies for sequence data and comparisons.

Key Insights and Ideas

  • Hox Cluster Duplication Leads to Massive Conservation Loss: The most striking finding is the significant reduction in conserved non-coding sequences (phylogenetic footprints) in Hox gene clusters after duplication events, observed in both zebrafish and pufferfish.
  • Beyond Gene Loss: While gene loss contributes to the loss of conservation, the observed reduction is greater than what can be explained by structural changes alone. This implies other evolutionary forces are at work.
  • Binding Site Turnover and Adaptive Modification: The excess loss of conservation suggests that binding site turnover (replacement of regulatory elements without functional change) and/or adaptive modification (changes driven by natural selection) play significant roles in shaping regulatory element evolution after duplication.
  • Differential Conservation in Paralog Clusters: In pufferfish, the two paralogous HoxA clusters exhibit notably different rates of non-structural sequence modification, with the HoxAb cluster showing a higher rate. This hints at potentially different evolutionary pressures or adaptive trajectories for these paralogs.
  • Tracker's Sensitivity: The developed "tracker" tool is significantly more sensitive than previous methods, identifying more footprints and enabling more comprehensive evolutionary analyses.

Key Results

  • The "tracker" software can survey Hox gene clusters in minutes, a substantial speedup compared to manual or web-based methods.
  • In zebrafish and pufferfish HoxA clusters, 70-90% of footprints found in shark or human are lost after duplication.
  • A probabilistic model for structural loss of PFCs (phylogenetic footprint clusters) was developed and applied.
  • The model predicted a minimal retention rate that was consistently higher than the observed rates, supporting the idea of non-structural loss mechanisms.
  • Estimates for non-structural PFC loss (â) were around 26-29% for zebrafish and pufferfish, with a higher rate (48%) observed for the pufferfish HoxAb cluster.

Limitations and Future Work

  • Incomplete Data: The analysis of other Hox clusters (Hox-C and Hox-D) is limited due to incomplete sequence data for duplicate clusters and the lack of suitable outgroup sequences.
  • PFC Definition: The definition of a phylogenetic footprint cluster (PFC) as a "single consistent clique" is pragmatic and might not capture all biologically relevant conserved regions, especially if multiple cliques are in close proximity.
  • Model Assumptions: The model for structural loss relies on assumptions about gene loss and cross-regulatory connectivity, and the degree of non-structural loss (a) is estimated rather than directly measured.
  • Future Work: The authors suggest that expression studies and transgenic tests could further investigate the adaptive modification hypothesis. They also highlight the need for more complete genomic data for other Hox clusters and the development of better models for non-coding sequence evolution. The "tracker" tool is intended for large-scale surveys, implying its applicability to other gene families and evolutionary scenarios.

Summary updated

BibTeX
@inproceedings{prohaska2003surveying,
  author = {Sonja J. Prohaska and Claudia Fried and Christoph Flamm and Günter P. Wagner and Peter F. Stadler},
  title = {{Surveying phylogenetic footprints in large gene clusters: applications to Hox cluster duplications}},
  booktitle = {Molecular},
  year = {2003},
  doi = {10.1016/j.ympev.2003.08.009},
}
Figure