Backtracking metabolic dynamics in single cells predicts bacterial replication in human macrophages

Area of Science:

• Cellular biology
• Infectious diseases
• Computational biology

Background:

• Tracking dynamic state transitions in cellular responses is vital for predicting biological outcomes.
• Understanding host-pathogen interactions requires detailed monitoring of cellular processes during infection.

Purpose of the Study:

• To develop a predictive model for bacterial infection in single human macrophages.
• To identify early metabolic markers associated with distinct infection outcomes, such as bacterial replication or cell death.

Main Methods:

• Live-cell imaging of human primary macrophages infected with Legionella pneumophila.
• Parallel monitoring of infection progression and cellular metabolic parameters.
• Development of a machine-learning model to predict infection outcomes.

Main Results:

• Early alterations in mitochondrial membrane potential (Δψm) and mitochondrial Reactive Oxygen Species (mROS) production were linked to subsequent bacterial growth.
• An explainable machine-learning model achieved 83% accuracy in predicting L. pneumophila replication prior to its onset.
• Backtracking analysis provided insights into host-pathogen dynamics and identified key predictive mitochondrial markers.

Conclusions:

• Early mitochondrial changes serve as critical predictive markers for successful bacterial infection.
• Machine learning models can accurately predict infection progression in single cells.
• This approach offers valuable insights into host-pathogen interactions and cellular response heterogeneity.