The γ-chain receptor dimerizes with complexes of the cytokines interleukin-2 (IL-2), IL-4, IL-7, IL-9, IL-15, and IL-21 and their corresponding "private" receptors. These cytokines have existing uses and future potential as immune therapies because of their ability to regulate the abundance and function of specific immune cell populations. Here, we build a binding reaction model for the ligand-receptor interactions of common γ-chain cytokines, which includes receptor trafficking dynamics, enabling quantitative predictions of cell-type-specific response to natural and engineered cytokines. We then show that tensor factorization is a powerful tool to visualize changes in the input-output behavior of the family across time, cell types, ligands, and concentrations. These results present a more accurate model of ligand response validated across a panel of immune cell types as well as a general approach for generating interpretable guidelines for manipulation of cell-type-specific targeting of engineered ligands.

Many receptor families exhibit both pleiotropy and redundancy in their regulation, with multiple ligands, receptors, and responding cell populations. Any intervention, therefore, has multiple effects, confounding intuition about how to precisely manipulate signaling for therapeutic purposes. The common γ-chain cytokine receptor dimerizes with complexes of the cytokines interleukin (IL)-2, IL-4, IL-7, IL-9, IL-15, and IL-21 and their corresponding “private” receptors. These cytokines have existing uses and future potential as immune therapies due to their ability to regulate the abundance and function of specific immune cell populations. However, engineering cell specificity into a therapy is confounded by the complexity of the family across responsive cell types. Here, we build a binding-reaction model for the ligand-receptor interactions of common γ-chain cytokines enabling quantitative predictions of response. We show that accounting for receptor-ligand trafficking is essential to accurately model cell response. This model accurately predicts ligand response across a wide panel of cell types under diverse experimental designs. Further, we can predict the effect and specificity of natural or engineered ligands across cell types. We then show that tensor factorization is a uniquely powerful tool to visualize changes in the input-output behavior of the family across time, cell types, ligands, and concentration. In total, these results present a more accurate model of ligand response validated across a panel of immune cell types, and demonstrate an approach for generating interpretable guidelines to manipulate the cell type-specific targeting of engineered ligands. These techniques will in turn help to study and therapeutically manipulate many other complex receptor-ligand families.

Summary points

A dynamical model of the γ-chain cytokines accurately models responses to IL-2, IL-15, IL-4, and IL-7. Receptor trafficking is necessary for capturing ligand response. Tensor factorization maps responses across cell populations, receptors, cytokines, and dynamics to visualize cytokine specificity. An activation model coupled with tensor factorization provides design specifications for engineering cell-specific responses.