A variational deep-learning approach to modeling memory T cell dynamics.

Area of Science:

• Immunology
• Computational Biology
• Systems Biology

Background:

• Mechanistic models of immune responses traditionally rely on simplified cell types.
• High-dimensional single-cell data presents challenges for existing modeling approaches.
• Understanding tissue-resident memory T cell dynamics is crucial for effective immunity.

Purpose of the Study:

• To develop a novel computational framework for analyzing high-dimensional single-cell data in immune responses.
• To investigate the dynamics and heterogeneity of lung tissue-resident memory CD4 and CD8 T cells during influenza infection resolution.
• To uncover mechanisms maintaining long-term immune memory in lung tissue.

Main Methods:

• Developed a simultaneous inference approach for dynamical model parameters and population structure.
• Utilized deep learning and stochastic variational inference trained on single-cell flow cytometry data.
• Applied the method to study CD4 and CD8 T cell populations in mice post-influenza infection.

Main Results:

• Identified significant phenotypic diversity within lung memory CD4 and CD8 T cells during immune response resolution.
• Demonstrated distinct, time-dependent dynamics among T cell subsets.
• Revealed that persistent Bcl-2hi subsets drive long-term heterogeneity and functional differentiation of memory T cells.

Conclusions:

• The developed computational approach provides new insights into tissue-localized immune memory dynamics.
• Phenotypic heterogeneity and specific differentiation pathways are key to maintaining long-term T cell memory.
• This method offers a novel basis for interpreting time-series high-dimensional biological data.