Systematic, multiparametric analysis of Mycobacterium tuberculosis intracellular infection offers insight into coordinated virulence

Abstract

A key to the pathogenic success of Mycobacterium tuberculosis (Mtb), the causative agent of tuberculosis, is the capacity to survive within host macrophages. Although several factors required for this survival have been identified, a comprehensive knowledge of such factors and how they work together to manipulate the host environment to benefit bacterial survival are not well understood. To systematically identify Mtb factors required for intracellular growth, we screened an arrayed, non-redundant Mtb transposon mutant library by high-content imaging to characterize the mutant-macrophage interaction. Based on a combination of imaging features, we identified mutants impaired for intracellular survival. We then characterized the phenotype of infection with each mutant by profiling the induced macrophage cytokine response. Taking a systems-level approach to understanding the biology of identified mutants, we performed a multiparametric analysis combining pathogen and host phenotypes to predict functional relationships between mutants based on clustering. Strikingly, mutants defective in two well-known virulence factors, the ESX-1 protein secretion system and the virulence lipid phthiocerol dimycocerosate (PDIM), clustered together. Building upon the shared phenotype of loss of the macrophage type I interferon (IFN) response to infection, we found that PDIM production and export are required for coordinated secretion of ESX-1-substrates, for phagosomal permeabilization, and for downstream induction of the type I IFN response. Multiparametric clustering also identified two novel genes that are required for PDIM production and induction of the type I IFN response. Thus, multiparametric analysis combining host and pathogen infection phenotypes can be used to identify novel functional relationships between genes that play a role in infection.

Path‐seq identifies an essential mycolate remodeling program for mycobacterial host adaptation

“…Interestingly, all of the PDIM‐related genes of module 276 genes ( ppsD/E , drrABC , papA5 ) were identified in a high‐content imaging analysis of bacterial mutants during macrophage infection (Barczak et al , ). More specifically, they found mutants of these genes impaired intracellular survival and reduced type I interferon (IFN) response in host cells (Barczak et al , ). Although the detrimental versus beneficial relevance of type I IFN in MTB infection remains a matter of active debate, growing evidence suggests type I IFN promotes bacterial expansion and pathogenesis within host cells (Moreira‐Teixeira et al , ).…”

“…The PDIM biosynthesis and transport genes, fad26, drrABC, and papA5, are additional gene members of module 276. Interestingly, all of the PDIM-related genes of module 276 genes (ppsD/E, drrABC, papA5) were identified in a highcontent imaging analysis of bacterial mutants during macrophage infection (Barczak et al, 2017). More specifically, they found mutants of these genes impaired intracellular survival and reduced type I interferon (IFN) response in host cells (Barczak et al, 2017 (Balhana et al, 2015).…”

“…Interestingly, all of the PDIM-related genes of module 276 genes (ppsD/E, drrABC, papA5) were identified in a highcontent imaging analysis of bacterial mutants during macrophage infection (Barczak et al, 2017). More specifically, they found mutants of these genes impaired intracellular survival and reduced type I interferon (IFN) response in host cells (Barczak et al, 2017 (Balhana et al, 2015). When overexpressed in MTB, Rv0472c led to significant repression of 15 genes, but only desA1 and desA2 had significant binding of Rv0472c in their promoter region from ChIP-seq analysis Data ref: Minch et al, 2015).…”

A novel deep learning scheme for morphology-based classification of mycobacterial infection in unstained macrophages

“…Unsupervised learning trains the predictive model with unlabeled data. For example, images features can be related to gene-mutation by unsupervised learning without any prior knowledge [24]. Supervised learning generally obtains more accurate results, owning to its abundant labeled data training process.…”

An essential mycolate remodeling program for mycobacterial adaptation in host cells

“…imaging of an MTB transposon mutant library (39). Leveraging our Path-seq data, we developed a systems-level approach that recapitulates known in vitro intracellular regulatory networks and prioritizes 234 others for further experimental testing.…”