Recombination produces coherent bacterial species clusters in both core and accessory genomes

Pekka Marttinen, Nicholas J. Croucher, Michael U. Gutmann, Jukka Corander, William P. HanageMicrobial 2015 · DOI

TL;DR

This paper introduces a novel computational model that simulates bacterial genome evolution, accounting for both core and accessory gene content. The model successfully explains the observed population structure of Streptococcus pneumoniae and suggests a data-driven approach to defining bacterial species based on genomic coherence.

What Problem Does This Paper Address?

Bacterial species are notoriously difficult to define due to their dynamic genomes, which include a stable "core" of essential genes and a variable "accessory" set acquired through horizontal gene transfer (HGT). While microbiologists agree that species should be "genomically coherent," there's no consensus on how to determine this, especially with the constant flux of genes. This paper tackles the challenge of understanding how both core and accessory genome variation arise and interact to shape bacterial populations, aiming to provide a more robust framework for species definition.

Methods

The authors developed a parsimonious computational model that simulates bacterial diversification. This model incorporates several key evolutionary processes:

  • Mutation: Standard genetic changes within genes.
  • Homologous Recombination (HR): The exchange of genetic material between similar DNA sequences, crucial for maintaining core genome integrity and shuffling accessory genes.
  • Horizontal Gene Transfer (HGT): The acquisition of new genes from unrelated organisms, contributing to the accessory genome.

A novel aspect of their approach is the simultaneous modeling of both core and accessory genome evolution, and importantly, the potential for HGT-introduced loci to be further shuffled by HR. The model uses a Wright-Fisher simulation framework and is neutral with respect to the fitness of individual genes or lineages. To prevent unlimited genome growth, a soft limit is imposed via a small multiplicative fitness penalty. Model fitting is achieved through simulation-based inference, matching summary statistics between simulated and real data, rather than traditional maximum-likelihood approaches which are computationally infeasible.

Datasets

The primary dataset used for evaluation is a collection of 616 Streptococcus pneumoniae genome sequences, previously described by Croucher et al. (2013). This dataset is characterized by its systematic sampling and comprehensive genomic information. The authors also provide supplementary animations and R code for their model, which can be considered a new benchmark for simulating bacterial population structure.

Key Insights and Ideas

  • Recombination as a Unifying Force: The model demonstrates that homologous recombination, acting on both core and accessory genomes, is a powerful driver in creating "coherent" bacterial clusters. It explains why strains with varying gene content can still form distinct, yet related, groups.
  • Equidistant Clusters: A surprising finding is that the model, without explicit selection, predicts a population structure where major strain clusters are roughly equidistant in terms of both core and accessory genome divergence. This suggests that the observed structure in S. pneumoniae is largely a consequence of neutral evolutionary processes and recombination dynamics.
  • A Data-Driven Species Definition: The model's ability to capture population structure provides a potential framework for defining bacterial species. Strains that deviate significantly from the model's predictions (i.e., are "genomically incoherent" with the rest of the population) could be rationally classified as distinct species.
  • Gene Content vs. Ecological Function: The model's neutrality regarding gene fitness suggests that simple comparisons of gene content might not be a reliable predictor of ecological function. The combination of genes, shaped by recombination, is more critical than individual gene presence or absence.

Key Results

  • The model successfully captures the major features of the S. pneumoniae population structure, including the characteristic U-shaped gene frequency histogram and the U-shaped distance distribution between strains.
  • The model estimates key evolutionary parameters (e.g., ratio of recombination to mutation, gene introduction/deletion rates) that are consistent with independent empirical estimates.
  • A clearly divergent subpopulation of pneumococci, previously identified as "atypical pneumococci," is identified as being inconsistent with the model, suggesting they may represent a distinct species.
  • Deviations from the model, such as a small peak near the origin in the distance distribution, are explained by factors like recent population bottlenecks or spatial structure, which warrant further investigation.

Limitations and Future Work

  • Idealized Model: The authors acknowledge that their model is idealized. For instance, the "soft limit" on genome size is a computational necessity and a simplification of complex biological realities.
  • Neutrality Assumption: While powerful for explaining population structure, the model's neutrality doesn't exclude the role of selection in shaping specific gene combinations or driving adaptation.
  • Distinguishing Barriers: The paper identifies a divergent cluster but cannot definitively determine the cause of the barrier to recombination (intrinsic vs. ecological).
  • Future Directions: The authors suggest that improved accessory genome annotation and extensions to their model could help investigate whether observed gene combinations are more or less frequent than expected by chance. Further work is needed to quantify the relative significance of bottlenecks and spatial structure in explaining observed deviations. The model also opens avenues for exploring the definition of "genomically coherent" species in other recombinogenic bacteria.

Summary updated

Abstract

The model does not include explicit selection on individual genes, suggesting that crude comparisons of gene content may be a poor predictor of ecological function. We identify a clearly divergent subpopulation of pneumococci that are inconsistent with the model and may be considered genomically incoherent with the rest of the population. These strains have a distinct disease tropism and may be rationally defined as a separate species. We also find deviations from the model that may be explained by recent population bottlenecks or spatial structure.
BibTeX
@inproceedings{marttinen2015recombination,
  author = {Pekka Marttinen and Nicholas J. Croucher and Michael U. Gutmann and Jukka Corander and William P. Hanage},
  title = {{Recombination produces coherent bacterial species clusters in both core and accessory genomes}},
  booktitle = {Microbial},
  year = {2015},
  doi = {10.1099/mgen.0.000038},
}

Loading related content...

Figure