Predicting effects of structural stress in a genome-reduced model bacterial metabolism

TL;DR

This study models how structural stress propagates in the metabolic network of Mycoplasma pneumoniae, revealing vulnerabilities, robustness features, and the impact of gene knockouts, with implications for understanding minimal bacterial genomes.

Contribution

It introduces a generic predictor for reaction vulnerability and analyzes non-linear cascade effects and gene-metabolism relationships in a genome-reduced bacterium.

Findings

Failure cascades vary between species and can be predicted by network motifs.

Genes controlling high-impact reactions are often functionally isolated.

M. pneumoniae exhibits evolved robustness despite its minimal genome.

Abstract

We studied in silico effects of structural stress in Mycoplasma pneumoniae, a genome-reduced model bacterial organism, by tracking the damage propagating on its metabolic network after a deleterious perturbation. First, we analyzed failure cascades spreading from individual reactions and pairs of reactions and compared the results to those in Staphylococcus aureus and Escherichia coli. To alert to the potential damage caused by the failure of individual reactions, we propose a generic predictor based on local information that identifies target reactions for structural vulnerability. With respect to the simultaneous failure of pairs of reactions, we detected strong non-linear amplification effects that can be predicted by the presence of specific motifs in the intersection of single cascades. We further connected the metabolic and gene co-expression networks of M. pneumoniae through enzyme activities, and studied the consequences of knocking out individual genes and clusters of genes. Damage caused by single gene knockouts reveals a strong correlation between genome-scale cascades of large impact and gene essentiality. At the same time, we found that genes controlling high-damage reactions tend to operate in functional isolation, as a metabolic protection mechanism. We conclude that the architecture of M. pneumoniae, both at the level of metabolism and genome, seems to have evolved towards increased structural robustness, similarly to other more complex model bacterial organisms, despite its reduced genome size and its greater metabolic network linearity. Our approach, although motivated biochemically, is generic enough to be of potential use toward analyzing and predicting spreading of structural stress in any bipartite complex network.