Mechanistic Computational Modeling of IL-2 Immunocytokines to Advance Immunotherapy

Introduction: Interleukin-2 (IL-2) stimulates the survival, activation, and expansion of T lymphocytes. Due to its critical role in immune function, the IL-2 cytokine has been FDA-approved for the treatment of certain metastatic cancers. However, the off-target effects of IL-2 and its vanishingly short half-life have hampered clinical progress. To circumvent the therapeutic shortcomings of the natural cytokine, our lab tethered IL-2 to anti-IL-2 antibodies to form immunocytokines, which enhance target specificity and significantly prolong the serum persistence of IL-2. To advance therapeutic translation, we are building a computational pharmacological model that mechanistically characterizes the activity of IL-2 and IL-2-based immunocytokines.

Materials and

Methods: We built and validated a computational mechanistic model that can simulate the dynamics of both IL-2 and IL-2 immunocytokines. The model incorporates ligand-receptor binding, trafficking dynamics, and signaling in two cell types (effector T cells and regulatory T cells). The model includes 12 molecules, 59 parameters, and 24 rate terms representing different mechanistic processes for the system, which were used to write 15 ordinary differential equations. The model was parameterized using experimental data when available, and physiological estimates when necessary. The level of IL-2 signaling induction (represented in the model as ligand-receptor binding) was used as a predictor of downstream intracellular signaling. The model was validated against experimental measurements of signaling (specifically, phosphorylation of signal transducer and activator of transcription 5 [STAT5]) induced by IL-2 and the IL-2-based immunocytokines.

Results, Conclusions, and Discussions: Based on initial model simulations, the relative difference in signal activation between the two T cell types following stimulation by IL-2 could be explained by the differences in receptor expression between those cell types; however, IL-2 signaling in the simulation based on ligand-receptor binding alone required IL-2 doses well above that needed to induce phosphorylation of STAT5 in the comparable experiments. We first hypothesized that binding rate constants might be faster than previously measured, and using parameter optimization, we were able to identify binding parameters that could reproduce the experimental data. However, this method was not parsimonious. Instead, we introduced an intermediate signaling step in the model between the ligand-receptor activation and the signaling readout, using a Hill function to permit amplification of the receptor activation. This transformation allowed for the previously measured binding affinities to reproduce the observed cell-type-specific responses to IL-2 using a consistent set of parameters. We used the same model to characterize how IL-2 differs from IL-2 immunocytokines in their activation of receptors and downstream signaling pathways. We particularly focused on bias of these IL-2-based molecules towards activation of various T cell subsets. We are now translating the mechanistic model to a computational pharmacological model to simulate IL-2 and the IL-2-based immunocytokine as therapeutics in the body, to help accelerate therapeutic regimen development. Further iteration between experimentation and modeling will make the computational mechanistic model more robust and predictive. Better mechanistic understanding of immunocytokines presents a new paradigm for the translation of safe and effective cytokine-based therapeutics, which has the potential to accelerate progress on the treatment of immune diseases.

Acknowledgements (Optional):