The pathobiology of Mycobacterium abscessus revealed through phenogenomic analysis

Lucas Boeck, Sophie Burbaud, Marcin J. Skwark, William H. Pearson, Jasper Sangen, Aaron Weimann, Isobel Everall, Josephine M. Bryant, Sony Malhotra, Bridget P. Bannerman, Katrin Kierdorf, Tom L. Blundell, Marc Dionne, Julian Parkhill, R. Andrés FlotobioRxiv 2021 · DOI

TL;DR

This study introduces a "phenogenomic" approach to dissect the complex pathobiology of Mycobacterium abscessus, a challenging pathogen. By integrating high-dimensional phenotypic data with whole-genome sequencing, computational modeling, and genetic interaction analysis, the researchers identified key gene networks controlling virulence and drug resistance, paving the way for new therapeutic targets.

What Problem Does This Paper Address?

Mycobacterium abscessus is a rapidly growing, multidrug-resistant bacterium that poses a significant threat, particularly to individuals with cystic fibrosis and other chronic lung diseases. Infections are difficult to treat, lead to severe lung damage, and can complicate lung transplantation. Despite its clinical importance, our understanding of how M. abscessus causes disease, how it develops antibiotic resistance, and what genetic factors drive its virulence remains limited. This knowledge gap hinders the development of effective treatments and strategies to predict patient outcomes. This paper aims to bridge this gap by providing a comprehensive, systems-level understanding of M. abscessus pathobiology.

Methods

The researchers employed a multi-pronged "phenogenomic" approach:

  • High-Dimensional Phenotyping: They characterized 331 clinical isolates of M. abscessus across 58 phenotypic dimensions. This included assessing planktonic growth, antibiotic resistance (both early and late), in vitro infection of human macrophage cell lines, in vivo infection of Drosophila melanogaster (measuring host survival and immune response), and correlating these with known clinical outcomes.
  • Genome-Wide Association Studies (GWAS): They performed GWAS for each phenotype to identify genetic variants associated with specific traits. This involved analyzing approximately 270,000 genetic variants (SNPs, insertions, deletions) using mixed models corrected for population structure.
  • Proteome-Wide Computational Structural Modelling: To interpret the functional impact of genetic variants and identify likely causal mutations, they used a graph-based machine learning method (mCSM) on a structural database of M. abscessus proteins. This helped assess how mutations might affect protein stability and function.
  • Direct Coupling Analysis (DCA): To uncover potential epistatic interactions (where the effect of one gene depends on the presence of another), they used ccDCA on whole-genome sequences from a large number of isolates. This method identifies co-evolving gene pairs that deviate from expected frequencies due to linkage disequilibrium, suggesting functional interaction.
  • In Vivo CRISPR-based Silencing: To experimentally validate key findings, they used inducible CRISPR interference (CRISPRi) in Drosophila melanogaster to silence specific genes and assess their impact on virulence.

Datasets

The study utilized:

  • 331 clinical isolates of Mycobacterium abscessus: These isolates were characterized phenotypically and their whole genomes were sequenced.
  • Previously collected metadata on patient clinical outcomes: This data was used to correlate bacterial phenotypes with disease progression.
  • A comprehensive M. abscessus structural database (Mabellini): Used for computational structural modeling.
  • Whole genome sequences from 2366 clinical isolates: Used for DCA analysis.

No new benchmarks were introduced; rather, existing methods were applied and integrated.

Key Insights and Ideas

  • Three Distinct Phenotypic Clusters: The study revealed three distinct groups of M. abscessus isolates, each associated with different clinical outcomes. This highlights that not all M. abscessus strains behave the same way and that subtle phenotypic differences can have significant clinical implications.
  • Beyond Macrolide Resistance: While macrolide resistance is a known factor, the study showed that even among macrolide-resistant isolates, clinical outcomes can vary dramatically, underscoring the importance of other, less understood, phenotypic characteristics in determining prognosis.
  • Identifying Causal Variants with Structural Modeling: Combining GWAS with computational structural modeling was crucial for pinpointing likely causal mutations. For example, a specific missense mutation in the mbtD gene was predicted to impair iron scavenging, directly linking a genetic change to a potential virulence mechanism.
  • Uncovering Epistatic Gene Networks: DCA successfully identified networks of co-evolving genes, suggesting functional interactions and epistatic relationships. This is a powerful way to understand how genes work together to influence complex traits like virulence.
  • A Novel Virulence Determinant: The research identified a novel gene network, including a previously uncharacterized peptide synthetase and a type 2 secretion system, that plays a critical role in regulating the pathogenic potential of M. abscessus. This is a significant discovery with direct therapeutic implications.

Key Results

  • 1926 Genetic Hits: The study identified 1926 genetic associations (involving 1000 genes) across 46 phenotypes through GWAS, confirming known resistance mechanisms and uncovering new ones.
  • Validation of Novel Genes: Experimental validation using in vivo CRISPRi in Drosophila confirmed that deletions in a putative Type II secretion system (MAB_0471) and a non-ribosomal peptide synthetase (MAB_3317c) led to increased host survival, indicating these genes are crucial for M. abscessus virulence.
  • Functional Interaction of Genes: Silencing of MAB_0472, a gene identified within the same epistatic network as MAB_0471 and MAB_3317c, also resulted in increased Drosophila survival, demonstrating functional interaction among these genes.

Limitations and Future Work

  • Model Organism Limitations: While Drosophila melanogaster is a valuable model, it doesn't perfectly replicate all aspects of human mycobacterial infection. Future work could involve validating findings in mammalian models.
  • CRISPRi in Liquid Culture: The study noted that gene silencing had no effect on bacterial growth in liquid media, suggesting that the identified genes' roles are primarily in the context of host interaction or in vivo conditions.
  • Therapeutic Targeting: The identification of a novel gene network regulating virulence presents a promising avenue for developing new therapeutic targets. Future research will likely focus on developing drugs that specifically inhibit these pathways.
  • Broader Applicability: The authors suggest that this phenogenomic approach is broadly applicable to other poorly understood bacterial pathogens, offering a rapid method for identifying prognostic indicators and novel drug targets.

Summary updated

Abstract

ABSTRACT The medical and scientific response to emerging pathogens is often severely hampered by ignorance of the genetic determinants of virulence, drug resistance, and clinical outcomes that could be used to identify therapeutic drug targets and forecast patient trajectories 1–5 . Taking the newly emergent multidrug-resistant bacteria Mycobacterium abscessus as an example 6 , we show that combining high dimensional phenotyping with whole genome sequencing in a phenogenomic analysis can rapidly reveal actionable systems-level insights into bacterial pathobiology. Using in vitro and in vivo phenotyping, we discovered three distinct clusters of isolates, each associated with a different clinical outcome. We combined genome-wide association studies (GWAS) with proteome-wide computational structural modelling 7 to define likely causal variants, and employed direct coupling analysis (DCA) 8 to identify co-evolving, and therefore potentially epistatic, gene networks. We then used in vivo CRISPR-based silencing to validate our findings, defining a novel secretion system controlling virulence in M. abscessus , and illustrating how phenogenomics can reveal critical pathways within emerging pathogenic bacteria.
BibTeX
@misc{boeck2021pathobiology,
  author = {Lucas Boeck and Sophie Burbaud and Marcin J. Skwark and William H. Pearson and Jasper Sangen and Aaron Weimann and Isobel Everall and Josephine M. Bryant and Sony Malhotra and Bridget P. Bannerman and Katrin Kierdorf and Tom L. Blundell and Marc Dionne and Julian Parkhill and R. Andrés Floto},
  title = {{The pathobiology of <i>Mycobacterium abscessus</i> revealed through phenogenomic analysis}},
  year = {2021},
  doi = {10.1101/2021.10.18.464689},
  howpublished = {bioRxiv},
}
Figure